Consolidate audit findings and record open decisions

.gitignore

@@ -22,6 +22,9 @@
 go.work
 go.work.sum
 
+# Local ReleaseKit delivery credentials
+.releasekit/
+
 # env file
 .env
 
@@ -67,4 +70,3 @@ Icon
 .AppleDesktop
 Network Trash Folder
 Temporary Items
-.apdisk

AGENTS.md

@@ -1,2 +1,6 @@
-Please review `docs/development.md` for initial orientation in this repository
-and follow its task-specific reading guide.
+Start with `docs/development.md` for orientation and use its policy reading guide
+to select the policies applicable to the task. Do not read the entire suite by
+default.
+
+Read and follow `docs/policy/agents.md` before recommending, planning, or
+implementing repository changes.

COPYRIGHT

@@ -0,0 +1,23 @@
+PromptKit
+
+Copyright (C) 2026 Eric Rakestraw
+
+SPDX-License-Identifier: LGPL-3.0-or-later
+
+This library is free software: you can redistribute it and/or modify it
+under the terms of the GNU Lesser General Public License as published by
+the Free Software Foundation, either version 3 of the License, or
+(at your option) any later version.
+
+The GNU Lesser General Public License version 3 is provided in LICENSE.
+
+This library is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY; without even the implied warranty of
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+GNU Lesser General Public License for more details.
+
+Policy documents copied or adapted from Standard Repository Policies remain licensed
+under CC-BY-SA-4.0. See docs/oss/attribution.md for the affected material,
+upstream credit, and license links.
+
+Third-party code and assets are subject to their respective license terms.

README.md

@@ -4,11 +4,12 @@ Promptkit is a reusable Go library for preparing and executing prompt-defined
 LLM workflows. Its module path is:
 
 ```text
-gitea.maximumdirect.net/eric/promptkit
+go.maximumdirect.net/promptkit
 ```
 
 The root `promptkit` package provides the supported public engine. Consumers
-can configure filesystem or in-memory prompt, profile, and schema sources,
+can configure file, embedded, or programmatic backend, profile, and prompt
+definitions alongside filesystem or in-memory schemas,
 prepare requests without generation, run requests with the built-in
 OpenAI-compatible client, or inject their own model client and artifact reader.
 
@@ -20,12 +21,23 @@ Run the maintained offline preparation example from the repository root:
 go run ./examples/go-library/prepare
 ```
 
-It loads a repository-local prompt, supplies an in-memory profile and inline
-input, and prints deterministic preparation metadata without contacting a
-provider or requiring credentials. Read the
-[example source](examples/go-library/prepare/main.go), the
+It composes two explicitly selected synthetic collections, imports
+source-relative prompt content, supplies an inline input, and prepares through
+a scoped view. The example also demonstrates collection inventories and
+namespace-filtered discovery, without contacting a provider or requiring
+credentials.
+The example disables the bundled OpenRouter configuration to keep its
+inventories limited to its own offline fixtures.
+The [consumer guide](docs/consumers/pkg-promptkit.md) also shows a local backend,
+ordered file/embedded sources, programmatic mappings, and how to replace an
+engine when source files change. Applications adapting the prior keyed
+configuration or versioned-prompt format can consult the
+[configuration migration guide](docs/consumers/configuration-migration.md).
+See the [example source](examples/go-library/prepare/main.go), the
 [Go package consumer guide](docs/consumers/pkg-promptkit.md), and the
 [framework format reference](docs/formats.md) to build a consumer workflow.
+The consumer guide also covers explicit, live smoke diagnostics and interpreting
+their factual reports; they are separate from the offline quickstart.
 
 Contributors should start with the [development guide](docs/development.md).
 The [architecture policy](docs/policy/architecture.md) defines the library
@@ -33,10 +45,53 @@ boundary and constraints that framework work must preserve.
 
 ## Release Guidance
 
-Consumers upgrading from `v0.6.0` to `v0.7.0` should read the
-[v0.7.0 changelog and migration guide](docs/releases/v0.7.0.md).
+The [v0.18.0 release notes](docs/releases/v0.18.0.md) cover the default-on
+OpenRouter backend and profiles, recorded configuration normalization, and
+ConfigKit v1.1. The notes explain inventory changes and the bundle opt-out;
+the Go 1.27 minimum remains unchanged.
+
+The [v0.17.0 release notes](docs/releases/v0.17.0.md) cover the vanity module
+path, ConfigKit v1, identified resource lists, removal of prompt-version
+metadata, and the Go 1.27 minimum. This is a breaking pre-v1 migration.
+
+Consumers using the former Go module path should follow the
+[module-path migration guidance](docs/consumers/configuration-migration.md#update-go-imports).
+
+The [v0.16.0 release notes](docs/releases/v0.16.0.md) cover explicit smoke
+diagnostics, configuration collections, scoped engine views, and discovery.
+Upgrading consumers should also review the current identified-resource and
+ConfigKit v1 changes in the [migration guide](docs/consumers/configuration-migration.md).
+
+The [v0.15.0 release notes](docs/releases/v0.15.0.md) cover error-only
+validation of concrete requests with `Engine.ValidateRequest`.
+
+Consumers upgrading to `v0.12.0` should read the
+[v0.12.0 release notes](docs/releases/v0.12.0.md) for the intentional ConfigKit
+configuration and public construction API migration.
+
+The [v0.11.2 release notes](docs/releases/v0.11.2.md) cover policy alignment
+and copied-policy attribution updates.
+
+The [v0.11.1 release notes](docs/releases/v0.11.1.md) cover the MIT-licensed
+catalog updates and licensing documentation improvements.
+
+Consumers upgrading to `v0.11.0` should read the
+[v0.11.0 changelog and migration guide](docs/releases/v0.11.0.md), particularly
+the default retry behavior and license change.
+
+Consumers upgrading to the `v0.10.0` profile-policy transition
+should read the [v0.10.0 transition guide](docs/releases/v0.10.0.md).
+
+Consumers upgrading from `v0.8.0` to `v0.9.0` should read the
+[v0.9.0 changelog and migration guide](docs/releases/v0.9.0.md).
+
+Consumers upgrading from `v0.7.0` to `v0.8.0` can consult the
+[v0.8.0 changelog and migration guide](docs/releases/v0.8.0.md).
 
 Earlier adopters can consult the
+[v0.7.0 changelog and migration guide](docs/releases/v0.7.0.md).
+
+Consumers upgrading from `v0.5.0` to `v0.6.0` can consult the
 [v0.6.0 changelog and migration guide](docs/releases/v0.6.0.md).
 
 Consumers upgrading from `v0.4.0` to `v0.5.0` can consult the
@@ -56,4 +111,8 @@ Consumers moving from `v0.1.0` to `v0.2.0` should read the
 [Scriptorium](https://gitea.maximumdirect.net/eric/scriptorium) is the CLI and
 HTTP application built on Promptkit.
 
-Promptkit is licensed under the [GNU General Public License version 3](LICENSE).
+## License
+
+Promptkit is licensed under the [LGPL v3.0 or later](COPYRIGHT).
+
+[Open Source attribution notices](docs/oss/attribution.md) are available.

appended_messages_contract_test.go

@@ -0,0 +1,129 @@
+package promptkit_test
+
+import (
+	"context"
+	"errors"
+	"reflect"
+	"sync/atomic"
+	"testing"
+
+	"go.maximumdirect.net/promptkit"
+)
+
+func TestAppendedMessageRoleConstantsAreStrings(t *testing.T) {
+	var (
+		developer string = promptkit.RoleDeveloper
+		system    string = promptkit.RoleSystem
+		user      string = promptkit.RoleUser
+		assistant string = promptkit.RoleAssistant
+	)
+	if developer != "developer" || system != "system" || user != "user" || assistant != "assistant" {
+		t.Fatalf("unexpected role constants: %q %q %q %q", developer, system, user, assistant)
+	}
+}
+
+func TestAppendedMessagesComposeFrozenEffectivePrompt(t *testing.T) {
+	cacheControl := &promptkit.CacheControl{Type: promptkit.CacheControlEphemeral, TTL: "1h"}
+	appended := []promptkit.RenderedMessage{
+		{Role: " \tAsSiStAnT\n", Content: "previous response"},
+		{Role: promptkit.RoleUser, Content: "corrective request", CacheControl: cacheControl},
+	}
+	client := &fakeLLMClient{response: &promptkit.GenerateResponse{Content: "output"}}
+	engine := newSimpleEngine(t, "configured prompt", promptkit.WithLLMClient(client))
+	request := promptkit.RunRequest{PromptID: "prompt", AppendedMessages: appended}
+
+	plain, err := engine.Prepare(context.Background(), promptkit.RunRequest{PromptID: "prompt"})
+	if err != nil {
+		t.Fatalf("prepare plain request: %v", err)
+	}
+	prepared, err := engine.Prepare(context.Background(), request)
+	if err != nil {
+		t.Fatalf("prepare appended request: %v", err)
+	}
+	expected := []promptkit.RenderedMessage{
+		{Role: promptkit.RoleUser, Content: "configured prompt"},
+		{Role: promptkit.RoleAssistant, Content: "previous response"},
+		{Role: promptkit.RoleUser, Content: "corrective request", CacheControl: &promptkit.CacheControl{Type: promptkit.CacheControlEphemeral, TTL: "1h"}},
+	}
+	if !reflect.DeepEqual(prepared.Messages, expected) || prepared.PromptHash != plain.PromptHash || prepared.RenderedPromptHash == plain.RenderedPromptHash {
+		t.Fatalf("prepared effective prompt = %#v, plain = %#v", prepared, plain)
+	}
+
+	equivalent, err := engine.Prepare(context.Background(), promptkit.RunRequest{PromptID: "prompt", AppendedMessages: []promptkit.RenderedMessage{
+		{Role: promptkit.RoleAssistant, Content: "previous response"},
+		{Role: promptkit.RoleUser, Content: "corrective request", CacheControl: &promptkit.CacheControl{Type: promptkit.CacheControlEphemeral, TTL: "1h"}},
+	}})
+	if err != nil || equivalent.RenderedPromptHash != prepared.RenderedPromptHash {
+		t.Fatalf("normalized-equivalent request = (%#v, %v), want matching rendered hash %q", equivalent, err, prepared.RenderedPromptHash)
+	}
+
+	for _, variant := range []promptkit.RunRequest{
+		{PromptID: "prompt", AppendedMessages: []promptkit.RenderedMessage{{Role: promptkit.RoleAssistant, Content: "changed"}, expected[2]}},
+		{PromptID: "prompt", AppendedMessages: []promptkit.RenderedMessage{expected[2], expected[1]}},
+		{PromptID: "prompt", AppendedMessages: []promptkit.RenderedMessage{{Role: promptkit.RoleAssistant, Content: "previous response"}, {Role: promptkit.RoleUser, Content: "corrective request"}}},
+	} {
+		variantPrepared, prepareErr := engine.Prepare(context.Background(), variant)
+		if prepareErr != nil || variantPrepared.RenderedPromptHash == prepared.RenderedPromptHash || variantPrepared.PromptHash != prepared.PromptHash {
+			t.Fatalf("variant preparation = (%#v, %v)", variantPrepared, prepareErr)
+		}
+	}
+
+	result, err := engine.Run(context.Background(), request)
+	if err != nil || result.RenderedPromptHash != prepared.RenderedPromptHash || !reflect.DeepEqual(client.requests[0].Prompt.Messages, expected) {
+		t.Fatalf("run result = (%#v, %v), request = %#v", result, err, client.requests)
+	}
+
+	execution, err := engine.PrepareExecution(context.Background(), request)
+	if err != nil {
+		t.Fatalf("prepare execution: %v", err)
+	}
+	appended[0].Content = "changed caller content"
+	cacheControl.TTL = ""
+	firstDetails := execution.Details()
+	firstDetails.Messages[1].Content = "changed details content"
+	firstDetails.Messages[2].CacheControl.Type = "changed details cache"
+	secondDetails := execution.Details()
+	if !reflect.DeepEqual(secondDetails.Messages, expected) || secondDetails.RenderedPromptHash != prepared.RenderedPromptHash {
+		t.Fatalf("prepared execution details = %#v, want frozen %#v", secondDetails, expected)
+	}
+
+	result, err = engine.RunPrepared(context.Background(), execution)
+	if err != nil || result.RenderedPromptHash != prepared.RenderedPromptHash || !reflect.DeepEqual(client.requests[1].Prompt.Messages, expected) {
+		t.Fatalf("prepared result = (%#v, %v), requests = %#v", result, err, client.requests)
+	}
+}
+
+func TestInvalidAppendedMessagesFailBeforeModelWork(t *testing.T) {
+	var modelCalls atomic.Int64
+	engine, err := promptkit.NewEngine(promptkit.Config{}, promptkit.WithLLMClient(countingLLMClient{calls: &modelCalls}))
+	if err != nil {
+		t.Fatalf("construct engine: %v", err)
+	}
+	request := promptkit.RunRequest{
+		PromptID: "unreached",
+		AppendedMessages: []promptkit.RenderedMessage{{
+			Role:    "unsupported-role",
+			Content: "sensitive appended content",
+		}},
+	}
+
+	operations := []struct {
+		name string
+		run  func() error
+	}{
+		{name: "Prepare", run: func() error { _, err := engine.Prepare(context.Background(), request); return err }},
+		{name: "PrepareExecution", run: func() error { _, err := engine.PrepareExecution(context.Background(), request); return err }},
+		{name: "Run", run: func() error { _, err := engine.Run(context.Background(), request); return err }},
+	}
+	for _, operation := range operations {
+		t.Run(operation.name, func(t *testing.T) {
+			err := operation.run()
+			if !errors.Is(err, promptkit.ErrInvalidRequest) {
+				t.Fatalf("error = %v, want ErrInvalidRequest", err)
+			}
+		})
+	}
+	if modelCalls.Load() != 0 {
+		t.Fatalf("invalid appended message invoked the model %d times", modelCalls.Load())
+	}
+}

artifact_reader.go

@@ -4,8 +4,8 @@ import (
 	"context"
 	"errors"
 
-	artifactadapter "gitea.maximumdirect.net/eric/promptkit/internal/artifact"
-	"gitea.maximumdirect.net/eric/promptkit/internal/domain"
+	artifactadapter "go.maximumdirect.net/promptkit/internal/artifact"
+	"go.maximumdirect.net/promptkit/internal/domain"
 )
 
 var errNilArtifactReaderResponse = errors.New("artifact reader returned nil artifact without error")

artifact_reader_internal_test.go

@@ -4,7 +4,7 @@ import (
 	"context"
 	"testing"
 
-	"gitea.maximumdirect.net/eric/promptkit/internal/domain"
+	"go.maximumdirect.net/promptkit/internal/domain"
 )
 
 func TestPublicArtifactReaderAdapterCopiesBody(t *testing.T) {

assembled_workflow_benchmark_test.go

@@ -0,0 +1,237 @@
+package promptkit_test
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"strings"
+	"testing"
+	"testing/fstest"
+
+	"go.maximumdirect.net/configkit"
+	"go.maximumdirect.net/promptkit"
+)
+
+var (
+	assembledPreparedSink *promptkit.PreparedRun
+	assembledResultSink   *promptkit.RunResult
+	assembledDetailsSink  promptkit.PreparedRun
+	assembledHandleSink   *promptkit.PreparedExecution
+)
+
+type assembledWorkflowClient struct{ content string }
+
+func (client assembledWorkflowClient) Generate(
+	context.Context,
+	promptkit.GenerateRequest,
+) (*promptkit.GenerateResponse, error) {
+	return &promptkit.GenerateResponse{Content: client.content}, nil
+}
+
+type assembledWorkflowFixture struct {
+	engine          *promptkit.Engine
+	request         promptkit.RunRequest
+	expectedMessage string
+	response        string
+}
+
+func assembledWorkflowSchema(b *testing.B) []byte {
+	b.Helper()
+	properties := make(map[string]any, 256)
+	for index := range 256 {
+		properties[fmt.Sprintf("property-%03d", index)] = map[string]any{"type": "string"}
+	}
+	data, err := json.Marshal(map[string]any{
+		"$schema":              "https://json-schema.org/draft/2020-12/schema",
+		"type":                 "object",
+		"properties":           properties,
+		"additionalProperties": false,
+	})
+	if err != nil {
+		b.Fatal(err)
+	}
+	return data
+}
+
+func assembledWorkflowExtras(entries int) map[string]any {
+	providers := make(map[string]any, entries)
+	for index := range entries {
+		providers[fmt.Sprintf("provider-%03d", index)] = map[string]any{
+			"enabled": true,
+			"labels":  []any{fmt.Sprintf("tier-%d", index%4), fmt.Sprintf("region-%d", index%3)},
+			"limits":  map[string]any{"burst": index + 1, "period": "minute"},
+			"routes":  []any{map[string]any{"weight": index + 1, "fallback": true}},
+		}
+	}
+	return map[string]any{"provider_options": providers}
+}
+
+func newAssembledWorkflowFixture(b *testing.B, name string) assembledWorkflowFixture {
+	b.Helper()
+	message := "Hello {{.name}}"
+	response := "ok"
+	output := map[string]any{"format": "text", "validation_mode": "none"}
+	profile := map[string]any{"id": "profile", "endpoint": "http://example.test/v1", "model": "benchmark-model"}
+	options := []promptkit.Option{promptkit.WithLLMClient(assembledWorkflowClient{content: response})}
+	var expectedMessage string
+
+	switch name {
+	case "small":
+		expectedMessage = "Hello Ada"
+	case "large_message":
+		prefix := strings.Repeat("x", 256*1024)
+		message = prefix + "{{.name}}"
+		expectedMessage = prefix + "Ada"
+	case "mutable_payload":
+		message = `{"subject":"{{.name}}"}`
+		expectedMessage = `{"subject":"Ada"}`
+		response = "{}"
+		profile["extra_params"] = assembledWorkflowExtras(64)
+		output = map[string]any{
+			"format": "json", "validation_mode": "json_schema", "schema_path": "workflow-schema.json",
+		}
+		options[0] = promptkit.WithLLMClient(assembledWorkflowClient{content: response})
+		options = append(options, promptkit.WithSchemaFS(fstest.MapFS{
+			"workflow-schema.json": &fstest.MapFile{Data: assembledWorkflowSchema(b)},
+		}, "."))
+	default:
+		b.Fatalf("unknown assembled workflow fixture %q", name)
+	}
+
+	config := map[string]any{
+		"profiles": []any{profile},
+		"prompts": []any{map[string]any{"id": "workflow",
+			"default_profile": "profile",
+			"messages":        []any{map[string]any{"role": "user", "content": message}},
+			"output":          output,
+		}},
+	}
+	engine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{Main: &configkit.Input{
+		Source: configkit.Mapping("assembled workflow benchmark", config),
+	}}}, options...)
+	if err != nil {
+		b.Fatal(err)
+	}
+	request := promptkit.RunRequest{
+		PromptID: "workflow", Vars: map[string]string{"name": "Ada"},
+	}
+	return assembledWorkflowFixture{
+		engine: engine, request: request, expectedMessage: expectedMessage, response: response,
+	}
+}
+
+func verifyAssembledWorkflowFixture(b *testing.B, fixture assembledWorkflowFixture) {
+	b.Helper()
+	ctx := context.Background()
+	prepared, err := fixture.engine.Prepare(ctx, fixture.request)
+	if err != nil {
+		b.Fatal(err)
+	}
+	if prepared.PromptID != "workflow" || len(prepared.Messages) != 1 ||
+		prepared.Messages[0].Content != fixture.expectedMessage {
+		b.Fatalf("prepared fixture output is invalid: %+v", prepared)
+	}
+	handle, err := fixture.engine.PrepareExecution(ctx, fixture.request)
+	if err != nil {
+		b.Fatal(err)
+	}
+	details := handle.Details()
+	if details.PromptID != "workflow" || len(details.Messages) != 1 || details.Messages[0].Content != fixture.expectedMessage {
+		handle.Discard()
+		b.Fatalf("prepared execution details are invalid: %+v", details)
+	}
+	handle.Discard()
+	result, err := fixture.engine.Run(ctx, fixture.request)
+	if err != nil {
+		b.Fatal(err)
+	}
+	if result.RawOutput != fixture.response {
+		b.Fatalf("run output=%q, want %q", result.RawOutput, fixture.response)
+	}
+	if fixture.response == "{}" && result.Validation.Status != promptkit.ValidationPassed {
+		b.Fatalf("structured response validation=%+v, want passed", result.Validation)
+	}
+}
+
+func BenchmarkEnginePreparation(b *testing.B) {
+	for _, name := range []string{"small", "large_message", "mutable_payload"} {
+		b.Run(name, func(b *testing.B) {
+			fixture := newAssembledWorkflowFixture(b, name)
+			verifyAssembledWorkflowFixture(b, fixture)
+			ctx := context.Background()
+			b.Run("prepare", func(b *testing.B) {
+				b.ReportAllocs()
+				b.ResetTimer()
+				for b.Loop() {
+					prepared, err := fixture.engine.Prepare(ctx, fixture.request)
+					if err != nil {
+						b.Fatal(err)
+					}
+					assembledPreparedSink = prepared
+				}
+				b.StopTimer()
+				assembledPreparedSink = nil
+			})
+			b.Run("prepared_execution_creation_and_discard", func(b *testing.B) {
+				b.ReportAllocs()
+				b.ResetTimer()
+				for b.Loop() {
+					handle, err := fixture.engine.PrepareExecution(ctx, fixture.request)
+					if err != nil {
+						b.Fatal(err)
+					}
+					handle.Discard()
+					assembledHandleSink = handle
+				}
+				b.StopTimer()
+				assembledHandleSink = nil
+			})
+		})
+	}
+}
+
+func BenchmarkEngineRun(b *testing.B) {
+	for _, name := range []string{"small", "large_message", "mutable_payload"} {
+		b.Run(name, func(b *testing.B) {
+			fixture := newAssembledWorkflowFixture(b, name)
+			verifyAssembledWorkflowFixture(b, fixture)
+			ctx := context.Background()
+			b.ReportAllocs()
+			b.ResetTimer()
+			for b.Loop() {
+				result, err := fixture.engine.Run(ctx, fixture.request)
+				if err != nil {
+					b.Fatal(err)
+				}
+				assembledResultSink = result
+			}
+			b.StopTimer()
+			assembledResultSink = nil
+		})
+	}
+}
+
+func BenchmarkPreparedExecutionDetails(b *testing.B) {
+	for _, name := range []string{"small", "large_message", "mutable_payload"} {
+		b.Run(name, func(b *testing.B) {
+			fixture := newAssembledWorkflowFixture(b, name)
+			verifyAssembledWorkflowFixture(b, fixture)
+			handle, err := fixture.engine.PrepareExecution(context.Background(), fixture.request)
+			if err != nil {
+				b.Fatal(err)
+			}
+			defer handle.Discard()
+			initial := handle.Details()
+			if len(initial.Messages) != 1 || initial.Messages[0].Content != fixture.expectedMessage {
+				b.Fatalf("prepared details fixture is invalid: %+v", initial)
+			}
+			b.ReportAllocs()
+			b.ResetTimer()
+			for b.Loop() {
+				assembledDetailsSink = handle.Details()
+			}
+			b.StopTimer()
+			assembledDetailsSink = promptkit.PreparedRun{}
+		})
+	}
+}

backends.go

@@ -1,107 +1,134 @@
 package promptkit
 
 import (
-	"gitea.maximumdirect.net/eric/promptkit/internal/backend"
-	"gitea.maximumdirect.net/eric/promptkit/internal/domain"
+	"go.maximumdirect.net/configkit"
+	"go.maximumdirect.net/promptkit/internal/backend"
 )
 
-// BackendOpenRouter is the reserved ID of Promptkit's built-in OpenRouter
-// backend.
+// BackendOpenRouter is the conventional ID for an OpenRouter backend.
 const BackendOpenRouter = backend.OpenRouterID
 
-// BackendRakestrawHome is the reserved ID of Promptkit's built-in
-// Rakestrawhome backend.
+// BackendRakestrawHome is the conventional ID for a Rakestrawhome backend.
 const BackendRakestrawHome = backend.RakestrawHomeID
 
-// BackendLocal is the case-sensitive conventional ID used by [LocalBackend].
-// It is not a built-in or reserved backend and must be registered with
-// [WithBackend].
+// BackendLocal is a conventional ID for a caller-configured local backend.
 const BackendLocal = "local"
 
-// Backend configures one engine-scoped OpenAI-compatible backend.
-//
-// Backend has no stable JSON representation. Use keyed literals so additions
-// to this configuration value do not break source compatibility.
-type Backend struct {
-	// ID is the stable, case-sensitive registry key. NewEngine trims it and
-	// requires a non-blank value. Built-in backend IDs are reserved.
+// BackendConfig is a presence-aware convenience value for an identified backend
+// definition. Nil pointer fields are omitted, while non-nil pointers preserve
+// explicit zero or blank values. A nil ExtraParams map is omitted; a non-nil
+// empty map supplies an empty mapping. A nil SmokeTests pointer is omitted; a
+// non-nil pointer supplies a smoke_tests mapping. Use raw ConfigKit Mapping
+// inputs for explicit null. BackendConfig has no stable JSON representation.
+type BackendConfig struct {
+	// ID is the local backend identity within its resource list.
 	ID string
-	// Endpoint is the OpenAI-compatible base endpoint. NewEngine trims it and
-	// requires an absolute HTTP or HTTPS URL with a host and without user
-	// information, a query string, or a fragment. Paths are allowed.
-	Endpoint string
-	// APIKeyEnv optionally names an environment lookup source for an API key.
-	// NewEngine trims it and requires the portable form [A-Za-z_][A-Za-z0-9_]*.
-	// A direct RunRequest.APIKey takes precedence. When no usable credential is
-	// available, the built-in client omits Authorization; injected clients own
-	// their own credential-resolution behavior. Store only the name, never a
-	// credential value.
-	APIKeyEnv string
-	// ExtraParams contains backend-wide request defaults. Values must be
-	// JSON-compatible, finite, acyclic, and keyed by non-empty strings. Keys
-	// must not be model, session_id, messages, temperature, max_tokens, top_p,
-	// service_tier, reasoning_effort, or response_format. An empty map supplies
-	// no defaults. NewEngine deeply copies the map and rejects excessively deep
-	// or large values for safety.
-	ExtraParams map[string]any
-	// ConcurrencyLimit is the maximum number of simultaneous model-generation
-	// calls allowed for this backend within one Engine. Zero leaves the backend
-	// unlimited. A negative value makes NewEngine fail with ErrInvalidConfig.
-	ConcurrencyLimit int
-	// QueueCapacity controls how many additional Run or RunPrepared calls may
-	// be admitted beyond ConcurrencyLimit. Nil uses 1024 when ConcurrencyLimit
-	// is positive; a pointer uses its exact value, including zero. The pointed-to
-	// value must be non-negative, and QueueCapacity must be nil when
-	// ConcurrencyLimit is zero. Their sum must fit in an int. WithBackend copies
-	// the value and does not retain the pointer.
+	// Endpoint is the required absolute HTTP(S) base URL after all layers merge.
+	Endpoint *string
+	// APIKeyEnv optionally names an environment variable, not a credential.
+	APIKeyEnv *string
+	// ConcurrencyLimit is zero for unlimited generation; negative is invalid.
+	ConcurrencyLimit *int
+	// QueueCapacity defaults to 1024 for positive concurrency when omitted.
+	// An explicit value, including zero, requires positive concurrency.
 	QueueCapacity *int
+	// ExtraParams supplies JSON-compatible provider request defaults.
+	ExtraParams map[string]any
+	// SmokeTests supplies optional diagnostic defaults and HTTP probe settings.
+	SmokeTests *BackendSmokeTestConfig
 }
 
-// LocalBackend returns a caller-owned Backend for a conventional local
-// OpenAI-compatible endpoint. It sets ID to BackendLocal and copies endpoint
-// and concurrencyLimit into Endpoint and ConcurrencyLimit without
-// normalization or validation. APIKeyEnv, ExtraParams, and QueueCapacity keep
-// their zero values.
-//
-// LocalBackend does not read environment variables, register the value, or
-// mutate engine or package state. Supply the returned value through
-// [WithBackend]; [NewEngine] then applies the ordinary backend validation and
-// concurrency semantics, including default queue capacity for a positive
-// limit, unlimited behavior for zero, and ErrInvalidConfig for a negative
-// limit.
-func LocalBackend(endpoint string, concurrencyLimit int) Backend {
-	return Backend{
-		ID:               BackendLocal,
-		Endpoint:         endpoint,
-		ConcurrencyLimit: concurrencyLimit,
-	}
+// BackendSmokeTestConfig contains optional diagnostic settings associated with
+// a backend. Nil nested fields are omitted by BackendSource.
+type BackendSmokeTestConfig struct {
+	// DefaultProfile is the exact profile ID selected as the backend's
+	// diagnostic inference default. A supplied value must not be blank; the ID
+	// is resolved only when that default is selected.
+	DefaultProfile *string
+	// Reachability optionally selects a full HTTP destination for reachability
+	// diagnostics. A nil endpoint uses the effective backend endpoint.
+	Reachability *ReachabilityProbeConfig
+	// Authentication optionally declares a named authentication diagnostic.
+	// A supplied object requires Kind.
+	Authentication *AuthenticationProbeConfig
 }
 
-// WithBackend adds one Backend registration to the constructed Engine.
-//
-// Registrations accumulate in option order. Every normalized ID must be unique
-// across consumer registrations and built-ins; a duplicate or invalid
-// definition makes NewEngine fail with ErrInvalidConfig. Built-in IDs,
-// including [BackendOpenRouter] and [BackendRakestrawHome], cannot be
-// replaced. The immutable registration is scoped to the resulting Engine and
-// cannot be enumerated, replaced, removed, or mutated after construction.
-// WithBackend does not install package-global state.
-func WithBackend(backend Backend) Option {
-	queueCapacity := 0
-	queueCapacitySet := backend.QueueCapacity != nil
-	if queueCapacitySet {
-		queueCapacity = *backend.QueueCapacity
-	}
-	return optionFunc(func(options *engineOptions) error {
-		options.backends = append(options.backends, domain.Backend{
-			ID:               backend.ID,
-			Endpoint:         backend.Endpoint,
-			APIKeyEnv:        backend.APIKeyEnv,
-			ExtraParams:      backend.ExtraParams,
-			ConcurrencyLimit: backend.ConcurrencyLimit,
-			QueueCapacity:    queueCapacity,
-			QueueCapacitySet: queueCapacitySet,
-		})
-		return nil
-	})
+// ReachabilityProbeConfig configures an optional reachability destination.
+type ReachabilityProbeConfig struct {
+	// Endpoint is an absolute HTTP(S) URL, including any desired path or query.
+	// Nil selects the effective backend endpoint; an explicit blank value is
+	// invalid.
+	Endpoint *string
+}
+
+// AuthenticationProbeConfig configures an optional named authentication
+// diagnostic.
+type AuthenticationProbeConfig struct {
+	// Kind selects the authentication adapter. Supported values are "none" and
+	// "openrouter_key"; surrounding whitespace and case are normalized.
+	Kind *string
+	// Endpoint is an absolute HTTP(S) URL, including any desired path or query.
+	// Nil selects the adapter's backend-relative destination; an explicit blank
+	// value is invalid.
+	Endpoint *string
+}
+
+// BackendSource creates a labeled ConfigKit mapping containing backend records
+// in caller order. IDs are explicit and duplicates are left for loading to reject.
+// It encodes fields without validating them; NewEngine validates the final
+// effective configuration after all selected source layers are combined.
+// Mapping captures a copy of supported input values at creation; unsupported
+// values fail when the source is loaded.
+func BackendSource(label string, backends []BackendConfig) configkit.Source {
+	definitions := make([]any, 0, len(backends))
+	for _, config := range backends {
+		fields := map[string]any{"id": config.ID}
+		if config.Endpoint != nil {
+			fields["endpoint"] = *config.Endpoint
+		}
+		if config.APIKeyEnv != nil {
+			fields["api_key_env"] = *config.APIKeyEnv
+		}
+		if config.ConcurrencyLimit != nil {
+			fields["concurrency_limit"] = *config.ConcurrencyLimit
+		}
+		if config.QueueCapacity != nil {
+			fields["queue_capacity"] = *config.QueueCapacity
+		}
+		if config.ExtraParams != nil {
+			fields["extra_params"] = providerMapForMapping(config.ExtraParams)
+		}
+		if config.SmokeTests != nil {
+			smokeTests := make(map[string]any)
+			if config.SmokeTests.DefaultProfile != nil {
+				smokeTests["default_profile"] = *config.SmokeTests.DefaultProfile
+			}
+			if config.SmokeTests.Reachability != nil {
+				reachability := make(map[string]any)
+				if config.SmokeTests.Reachability.Endpoint != nil {
+					reachability["endpoint"] = *config.SmokeTests.Reachability.Endpoint
+				}
+				smokeTests["reachability"] = reachability
+			}
+			if config.SmokeTests.Authentication != nil {
+				authentication := make(map[string]any)
+				if config.SmokeTests.Authentication.Kind != nil {
+					authentication["kind"] = *config.SmokeTests.Authentication.Kind
+				}
+				if config.SmokeTests.Authentication.Endpoint != nil {
+					authentication["endpoint"] = *config.SmokeTests.Authentication.Endpoint
+				}
+				smokeTests["authentication"] = authentication
+			}
+			fields["smoke_tests"] = smokeTests
+		}
+		definitions = append(definitions, fields)
+	}
+	return configkit.Mapping(label, map[string]any{"backends": definitions})
+}
+
+// LocalBackend supplies endpoint and concurrency pointers for a caller-chosen
+// backend ID; it does not select an ID or validate either value.
+func LocalBackend(endpoint string, concurrencyLimit int) BackendConfig {
+	return BackendConfig{Endpoint: &endpoint, ConcurrencyLimit: &concurrencyLimit}
 }

bundled_composition_contract_test.go

@@ -0,0 +1,115 @@
+package promptkit_test
+
+import (
+	"context"
+	"fmt"
+	"strings"
+	"testing"
+
+	"go.maximumdirect.net/configkit"
+	"go.maximumdirect.net/promptkit"
+)
+
+func TestBundledCompositionReplacementAndInheritance(t *testing.T) {
+	underlay := configkit.Input{Source: configkit.Mapping("caller underlay", map[string]any{
+		"profiles": []any{map[string]any{"id": "gpt-sol-latest", "model": "underlay-model"}},
+	})}
+	underlayEngine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{Underlays: []configkit.Underlay{{Inputs: []configkit.Input{underlay}}}}})
+	if err != nil {
+		t.Fatal(err)
+	}
+	underlayProfile, err := underlayEngine.InspectProfile(context.Background(), "gpt-sol-latest")
+	if err != nil || underlayProfile.EffectiveModelParams.Model != "underlay-model" || underlayProfile.EffectiveModelParams.BackendID != "" {
+		t.Fatalf("underlay replacement = %+v, %v", underlayProfile, err)
+	}
+	main := configkit.Input{Source: configkit.Mapping("caller main", map[string]any{
+		"backends": []any{map[string]any{"id": "openrouter", "endpoint": "https://local.invalid/v1"}, map[string]any{"id": "other", "endpoint": "https://other.invalid/v1"}},
+		"profiles": []any{
+			map[string]any{"id": "gpt-sol-latest", "model": "main-model"},
+			map[string]any{"id": "custom", "extends": map[string]any{"id": "gpt-luna-latest"}, "model": "custom-model"},
+			map[string]any{"id": "auto", "backend": "other", "model": "other-model"},
+		},
+		"collections": map[string]any{"openrouter": map[string]any{
+			"backends": []any{map[string]any{"id": "local", "endpoint": "https://collection.invalid/v1"}},
+			"profiles": []any{map[string]any{"id": "profile", "backend": "openrouter/local", "model": "collection-model"}},
+		}},
+	})}
+	engine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{Underlays: []configkit.Underlay{{Inputs: []configkit.Input{underlay}}}, Main: &main}})
+	if err != nil {
+		t.Fatal(err)
+	}
+	inspect := func(id string) promptkit.ExecutionTarget {
+		t.Helper()
+		profile, err := engine.InspectProfile(context.Background(), id)
+		if err != nil {
+			t.Fatal(err)
+		}
+		return profile.EffectiveModelParams
+	}
+	if target := inspect("gpt-sol-latest"); target.Model != "main-model" || target.BackendID != "" {
+		t.Fatalf("replacement = %+v", target)
+	}
+	if target := inspect("custom"); target.Model != "custom-model" || target.BackendID != "openrouter" || target.Endpoint != "https://local.invalid/v1" {
+		t.Fatalf("inherited = %+v", target)
+	}
+	if target := inspect("gpt-luna-latest"); target.Endpoint != "https://local.invalid/v1" || target.APIKeyEnv != "" {
+		t.Fatalf("backend replacement = %+v", target)
+	}
+	if target := inspect("openrouter/profile"); target.BackendID != "openrouter/local" || target.Model != "collection-model" {
+		t.Fatalf("collection = %+v", target)
+	}
+	for _, tier := range []string{"low", "medium", "high", "xhigh", "max"} {
+		target := inspect("auto-" + tier)
+		if target.BackendID != "other" || target.Model != "other-model" || !strings.Contains(fmt.Sprint(target.ExtraParams), "cost_tier:"+tier) || !strings.Contains(fmt.Sprint(target.ExtraParams), "id:auto-router") {
+			t.Errorf("tier %s = %+v", tier, target)
+		}
+	}
+}
+
+func TestBundledCompositionSequenceModes(t *testing.T) {
+	for _, test := range []struct {
+		name, mode, id string
+		wantErr        bool
+		count          int
+	}{
+		{"duplicate append", "append", "auto", true, 0},
+		{"distinct append", "append", "custom", false, 18},
+		{"replace", "replace", "custom", false, 1},
+	} {
+		t.Run(test.name, func(t *testing.T) {
+			input := configkit.Input{Source: configkit.Mapping("sequence", map[string]any{
+				"config":   map[string]any{"sequences": []any{map[string]any{"destination": []any{"profiles"}, "mode": test.mode}}},
+				"profiles": []any{map[string]any{"id": test.id, "model": "custom-model"}},
+			})}
+			engine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{Main: &input}})
+			if test.wantErr {
+				if err == nil {
+					t.Fatal("expected duplicate rejection")
+				}
+				return
+			}
+			if err != nil {
+				t.Fatal(err)
+			}
+			profiles, err := engine.ListProfiles(context.Background())
+			if err != nil || len(profiles) != test.count {
+				t.Fatalf("profiles = %v, %v", profiles, err)
+			}
+		})
+	}
+}
+
+func TestBundledCompositionRejectsStrictConflictsAndMissingBackends(t *testing.T) {
+	profile := configkit.Input{Source: configkit.Mapping("profile", map[string]any{"profiles": []any{map[string]any{"id": "auto", "model": "caller"}}})}
+	_, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{Underlays: []configkit.Underlay{{Inputs: []configkit.Input{profile, profile}}}}})
+	if err == nil {
+		t.Fatal("strict group accepted duplicate records")
+	}
+	input := configkit.Input{Source: configkit.Mapping("replace backends", map[string]any{
+		"config":   map[string]any{"sequences": []any{map[string]any{"destination": []any{"backends"}, "mode": "replace"}}},
+		"backends": []any{map[string]any{"id": "other", "endpoint": "https://other.invalid/v1"}},
+	})}
+	if _, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{Main: &input}}); err == nil {
+		t.Fatal("missing bundled backend was accepted")
+	}
+}

bundled_execution_contract_test.go

@@ -0,0 +1,90 @@
+package promptkit_test
+
+import (
+	"context"
+	"encoding/json"
+	"errors"
+	"io"
+	"net/http"
+	"reflect"
+	"strings"
+	"testing"
+
+	"go.maximumdirect.net/configkit"
+	"go.maximumdirect.net/promptkit"
+)
+
+func bundledPromptConfig(client *http.Client) promptkit.Config {
+	input := configkit.Input{Source: configkit.Mapping("prompt", map[string]any{"prompts": []any{map[string]any{
+		"id": "prompt", "messages": []any{map[string]any{"role": "user", "content": "hello"}}, "output": map[string]any{"format": "text", "validation_mode": "none"},
+	}}})}
+	return promptkit.Config{HTTPClient: client, Sources: promptkit.ConfigurationSources{Main: &input}}
+}
+
+func TestBundledProfilePreparationAndExecution(t *testing.T) {
+	client := &fakeLLMClient{response: &promptkit.GenerateResponse{Content: "ok"}}
+	engine, err := promptkit.NewEngine(bundledPromptConfig(nil), promptkit.WithLLMClient(client))
+	if err != nil {
+		t.Fatal(err)
+	}
+	if _, err := engine.PrepareExecution(context.Background(), promptkit.RunRequest{PromptID: "prompt"}); !errors.Is(err, promptkit.ErrProfileRequired) {
+		t.Fatalf("missing profile = %v", err)
+	}
+	temperature := 0.3
+	request := promptkit.RunRequest{PromptID: "prompt", ProfileID: "gpt-luna-latest", Execution: &promptkit.ExecutionTargetOverride{Temperature: &temperature}}
+	prepared, err := engine.PrepareExecution(context.Background(), request)
+	if err != nil {
+		t.Fatal(err)
+	}
+	details := prepared.Details()
+	if details.SelectedProfileID != "gpt-luna-latest" || details.EffectiveModelParams.Model != "~openai/gpt-luna-latest" || details.EffectiveModelParams.BackendID != "openrouter" || details.EffectiveModelParams.Temperature != temperature {
+		t.Fatalf("prepared target = %+v", details.EffectiveModelParams)
+	}
+	if _, err := engine.Run(context.Background(), request); err != nil {
+		t.Fatal(err)
+	}
+	if len(client.requests) != 1 || client.requests[0].Target.Model != "~openai/gpt-luna-latest" || client.requests[0].Target.BackendID != "openrouter" {
+		t.Fatalf("requests = %+v", client.requests)
+	}
+}
+
+func TestBundledProfileHTTPPayloads(t *testing.T) {
+	t.Setenv("OPENROUTER_API_KEY", "synthetic-key")
+	for _, test := range []struct{ id, model, tier string }{
+		{"auto-low", "openrouter/auto", "low"}, {"auto-medium", "openrouter/auto", "medium"}, {"auto-high", "openrouter/auto", "high"}, {"auto-xhigh", "openrouter/auto", "xhigh"}, {"auto-max", "openrouter/auto", "max"},
+		{"auto", "openrouter/auto", ""}, {"free", "openrouter/free", ""}, {"gpt-luna-latest", "~openai/gpt-luna-latest", ""},
+	} {
+		t.Run(test.id, func(t *testing.T) {
+			var body map[string]any
+			transport := roundTripFunc(func(request *http.Request) (*http.Response, error) {
+				if request.Header.Get("Authorization") != "Bearer synthetic-key" {
+					t.Errorf("authorization = %q", request.Header.Get("Authorization"))
+				}
+				if err := json.NewDecoder(request.Body).Decode(&body); err != nil {
+					t.Fatal(err)
+				}
+				return &http.Response{StatusCode: 200, Header: make(http.Header), Body: io.NopCloser(strings.NewReader(`{"choices":[{"message":{"content":"ok"}}]}`))}, nil
+			})
+			engine, err := promptkit.NewEngine(bundledPromptConfig(&http.Client{Transport: transport}))
+			if err != nil {
+				t.Fatal(err)
+			}
+			if _, err := engine.Run(context.Background(), promptkit.RunRequest{PromptID: "prompt", ProfileID: test.id}); err != nil {
+				t.Fatal(err)
+			}
+			if body["model"] != test.model || body["extra_params"] != nil {
+				t.Fatalf("request = %+v", body)
+			}
+			if test.tier == "" {
+				if _, exists := body["plugins"]; exists {
+					t.Fatalf("unexpected plugins = %+v", body)
+				}
+				return
+			}
+			want := []any{map[string]any{"id": "auto-router", "cost_tier": test.tier}}
+			if !reflect.DeepEqual(body["plugins"], want) {
+				t.Fatalf("plugins = %#v, want %#v", body["plugins"], want)
+			}
+		})
+	}
+}

bundled_openrouter_contract_test.go

@@ -0,0 +1,124 @@
+package promptkit_test
+
+import (
+	"context"
+	"errors"
+	"net/http"
+	"reflect"
+	"strings"
+	"testing"
+
+	"go.maximumdirect.net/configkit"
+	"go.maximumdirect.net/promptkit"
+)
+
+type rejectBundleTransport struct{}
+
+func (rejectBundleTransport) RoundTrip(*http.Request) (*http.Response, error) {
+	return nil, errors.New("unexpected HTTP request")
+}
+
+func TestBundledOpenRouterOptionsAndInventory(t *testing.T) {
+	t.Setenv("OPENROUTER_API_KEY", "")
+	for _, test := range []struct {
+		name    string
+		options []promptkit.Option
+		enabled bool
+	}{
+		{"default", nil, true}, {"enabled", []promptkit.Option{promptkit.WithBundledOpenRouter(true)}, true},
+		{"disabled", []promptkit.Option{promptkit.WithBundledOpenRouter(false)}, false},
+		{"last disables", []promptkit.Option{promptkit.WithBundledOpenRouter(true), promptkit.WithBundledOpenRouter(false)}, false},
+		{"last enables", []promptkit.Option{promptkit.WithBundledOpenRouter(false), promptkit.WithBundledOpenRouter(true)}, true},
+	} {
+		t.Run(test.name, func(t *testing.T) {
+			engine, err := promptkit.NewEngine(promptkit.Config{HTTPClient: &http.Client{Transport: rejectBundleTransport{}}}, test.options...)
+			if err != nil {
+				t.Fatal(err)
+			}
+			profiles, err := engine.ListProfiles(context.Background())
+			if err != nil {
+				t.Fatal(err)
+			}
+			backends, err := engine.ListBackends(context.Background())
+			if err != nil {
+				t.Fatal(err)
+			}
+			prompts, err := engine.ListPrompts(context.Background())
+			if err != nil {
+				t.Fatal(err)
+			}
+			collections, err := engine.ListCollections(context.Background())
+			if err != nil {
+				t.Fatal(err)
+			}
+			want := 0
+			if test.enabled {
+				want = 17
+			}
+			if len(profiles) != want || len(backends) != boolCount(test.enabled) || len(prompts) != 0 || len(collections) != 0 {
+				t.Fatalf("inventories: %v %v %v %v", profiles, backends, prompts, collections)
+			}
+			if test.enabled {
+				wantProfiles := []string{"auto", "auto-high", "auto-low", "auto-max", "auto-medium", "auto-xhigh", "claude-fable-latest", "claude-haiku-latest", "claude-opus-latest", "claude-sonnet-latest", "free", "gemini-flash-latest", "gemini-pro-latest", "gpt-astra-latest", "gpt-luna-latest", "gpt-sol-latest", "gpt-terra-latest"}
+				if !reflect.DeepEqual(profiles, wantProfiles) || !reflect.DeepEqual(backends, []string{"openrouter"}) {
+					t.Fatalf("bundle inventory: %v %v", profiles, backends)
+				}
+				for id, model := range map[string]string{"auto": "openrouter/auto", "auto-max": "openrouter/auto", "gpt-luna-latest": "~openai/gpt-luna-latest", "free": "openrouter/free"} {
+					profile, err := engine.InspectProfile(context.Background(), id)
+					if err != nil || profile.EffectiveModelParams.Model != model || profile.EffectiveModelParams.BackendID != "openrouter" {
+						t.Fatalf("profile %s = %+v, %v", id, profile, err)
+					}
+				}
+				view, err := engine.InspectConfiguration()
+				if err != nil {
+					t.Fatal(err)
+				}
+				if !strings.Contains(fmtSources(view.Sources()), "Promptkit bundled OpenRouter") {
+					t.Fatal("bundle source missing")
+				}
+				if explanation, found := view.Explain(configkit.Keys("backends").Index(0).Key("endpoint")); !found || !strings.Contains(explanation.Effective.YAML, "https://openrouter.ai/api/v1") || len(explanation.History) == 0 {
+					t.Fatalf("bundle explanation = %+v", explanation)
+				}
+			}
+		})
+	}
+}
+
+func boolCount(value bool) int {
+	if value {
+		return 1
+	}
+	return 0
+}
+func fmtSources(sources []configkit.SourceInfo) string {
+	var b strings.Builder
+	for _, source := range sources {
+		b.WriteString(source.Label)
+	}
+	return b.String()
+}
+
+func TestBundlePreservesCallerUnderlayBackingArray(t *testing.T) {
+	source := configkit.Mapping("caller", map[string]any{"profiles": []any{map[string]any{"id": "caller", "model": "model"}}})
+	groups := make([]configkit.Underlay, 1, 2)
+	groups[0] = configkit.Underlay{Inputs: []configkit.Input{{Source: source}}}
+	before := append([]configkit.Underlay(nil), groups...)
+	if _, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{Underlays: groups}}); err != nil {
+		t.Fatal(err)
+	}
+	if !reflect.DeepEqual(groups, before) || !reflect.DeepEqual(groups[:2][1], configkit.Underlay{}) {
+		t.Fatal("caller underlays were modified")
+	}
+}
+
+func TestDisabledBundleKeepsCallerOpenRouter(t *testing.T) {
+	input := configkit.Input{Source: configkit.Mapping("caller", map[string]any{"backends": []any{map[string]any{"id": "openrouter", "endpoint": "https://local.invalid/v1"}}})}
+	engine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{Main: &input}}, promptkit.WithBundledOpenRouter(false))
+	if err != nil {
+		t.Fatal(err)
+	}
+	backends, err := engine.ListBackends(context.Background())
+	if err != nil || !reflect.DeepEqual(backends, []string{"openrouter"}) {
+		t.Fatalf("backends = %v, %v", backends, err)
+	}
+}

bundled_smoke_contract_test.go

@@ -0,0 +1,127 @@
+package promptkit_test
+
+import (
+	"context"
+	"encoding/json"
+	"io"
+	"net/http"
+	"reflect"
+	"strings"
+	"testing"
+
+	"go.maximumdirect.net/configkit"
+	"go.maximumdirect.net/promptkit"
+)
+
+func TestBundledSmokeHTTPDefaults(t *testing.T) {
+	t.Setenv("OPENROUTER_API_KEY", "synthetic-key")
+	var urls []string
+	var models []string
+	transport := roundTripFunc(func(request *http.Request) (*http.Response, error) {
+		urls = append(urls, request.URL.String())
+		body := `{"data":{"label":"member","is_free_tier":false,"usage":0}}`
+		if request.Method == http.MethodPost {
+			var payload map[string]any
+			if err := json.NewDecoder(request.Body).Decode(&payload); err != nil {
+				t.Fatal(err)
+			}
+			models = append(models, payload["model"].(string))
+			content := "PROMPTKIT_OK"
+			if _, structured := payload["response_format"]; structured {
+				content = `{"status":"ok"}`
+			}
+			encoded, err := json.Marshal(map[string]any{"choices": []any{map[string]any{"message": map[string]any{"content": content}}}})
+			if err != nil {
+				t.Fatal(err)
+			}
+			body = string(encoded)
+		} else if strings.HasSuffix(request.URL.Path, "/key") && request.Header.Get("Authorization") != "Bearer synthetic-key" {
+			t.Errorf("authentication header missing")
+		}
+		return &http.Response{StatusCode: 200, Header: make(http.Header), Body: io.NopCloser(strings.NewReader(body))}, nil
+	})
+	engine, err := promptkit.NewEngine(promptkit.Config{HTTPClient: &http.Client{Transport: transport}})
+	if err != nil {
+		t.Fatal(err)
+	}
+	before, _ := engine.ListProfiles(context.Background())
+	inspectionBefore, err := engine.InspectConfiguration()
+	if err != nil {
+		t.Fatal(err)
+	}
+	for _, kind := range []promptkit.SmokeTestKind{promptkit.SmokeTestReachability, promptkit.SmokeTestAuthentication, promptkit.SmokeTestTextInference, promptkit.SmokeTestStructuredOutput} {
+		report, err := engine.SmokeTest(context.Background(), promptkit.SmokeTestRequest{Kind: kind, BackendID: "openrouter"})
+		if err != nil || report == nil || !report.Completed || report.Failure != "" {
+			t.Fatalf("%s report = %+v, %v", kind, report, err)
+		}
+		if kind == promptkit.SmokeTestAuthentication && (report.Facts.CredentialsAccepted == nil || !*report.Facts.CredentialsAccepted) {
+			t.Fatalf("authentication facts = %+v", report.Facts)
+		}
+		if kind == promptkit.SmokeTestTextInference || kind == promptkit.SmokeTestStructuredOutput {
+			if report.Target.ProfileID != "gpt-luna-latest" || report.Target.EffectiveSettings.Model != "~openai/gpt-luna-latest" {
+				t.Fatalf("target = %+v", report.Target)
+			}
+		}
+	}
+	wantURLs := []string{"https://openrouter.ai/api/v1", "https://openrouter.ai/api/v1/key", "https://openrouter.ai/api/v1/chat/completions", "https://openrouter.ai/api/v1/chat/completions"}
+	if !reflect.DeepEqual(urls, wantURLs) || !reflect.DeepEqual(models, []string{"~openai/gpt-luna-latest", "~openai/gpt-luna-latest"}) {
+		t.Fatalf("urls %v, models %v", urls, models)
+	}
+	after, _ := engine.ListProfiles(context.Background())
+	if !reflect.DeepEqual(before, after) {
+		t.Fatalf("inventory changed: %v %v", before, after)
+	}
+	inspectionAfter, err := engine.InspectConfiguration()
+	if err != nil {
+		t.Fatal(err)
+	}
+	diff, err := inspectionBefore.Diff(inspectionAfter)
+	if err != nil || len(diff) != 0 {
+		t.Fatalf("inspection changed: %+v, %v", diff, err)
+	}
+}
+
+func TestBundledSmokeSelectionUsesEffectiveConfiguration(t *testing.T) {
+	replacement := configkit.Input{Source: configkit.Mapping("replacement", map[string]any{"profiles": []any{map[string]any{"id": "gpt-luna-latest", "backend": "openrouter", "model": "replacement"}}})}
+	replacementClient := &fakeLLMClient{response: &promptkit.GenerateResponse{Content: "PROMPTKIT_OK"}}
+	replacementEngine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{Main: &replacement}}, promptkit.WithLLMClient(replacementClient))
+	if err != nil {
+		t.Fatal(err)
+	}
+	replacementReport, err := replacementEngine.SmokeTest(context.Background(), promptkit.SmokeTestRequest{Kind: promptkit.SmokeTestTextInference, BackendID: "openrouter"})
+	if err != nil || replacementReport.Target.ProfileID != "gpt-luna-latest" || replacementReport.Target.EffectiveSettings.Model != "replacement" {
+		t.Fatalf("replaced default = %+v, %v", replacementReport, err)
+	}
+	replacementClient.response.Content = "unexpected"
+	mismatch, err := replacementEngine.SmokeTest(context.Background(), promptkit.SmokeTestRequest{Kind: promptkit.SmokeTestTextInference, BackendID: "openrouter"})
+	if err != nil || mismatch == nil || !mismatch.Completed {
+		t.Fatalf("concluded mismatch = %+v, %v", mismatch, err)
+	}
+	input := configkit.Input{Source: configkit.Mapping("application", map[string]any{
+		"backends": []any{map[string]any{"id": "openrouter", "endpoint": "https://example.invalid/v1", "smoke_tests": map[string]any{"default_profile": "custom"}}},
+		"profiles": []any{map[string]any{"id": "gpt-luna-latest", "backend": "openrouter", "model": "replacement"}, map[string]any{"id": "custom", "backend": "openrouter", "model": "custom-model"}},
+		"prompts":  []any{map[string]any{"id": "prompt", "default_profile": "free", "messages": []any{map[string]any{"role": "user", "content": "private application marker"}}, "output": map[string]any{"format": "text", "validation_mode": "none"}}},
+	})}
+	client := &fakeLLMClient{response: &promptkit.GenerateResponse{Content: "PROMPTKIT_OK"}}
+	engine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{Main: &input}}, promptkit.WithLLMClient(client))
+	if err != nil {
+		t.Fatal(err)
+	}
+	for _, test := range []struct {
+		request        promptkit.SmokeTestRequest
+		profile, model string
+	}{
+		{promptkit.SmokeTestRequest{BackendID: "openrouter"}, "custom", "custom-model"},
+		{promptkit.SmokeTestRequest{BackendID: "openrouter", PromptID: "prompt"}, "free", "openrouter/free"},
+		{promptkit.SmokeTestRequest{BackendID: "openrouter", PromptID: "prompt", ProfileID: "gpt-luna-latest"}, "gpt-luna-latest", "replacement"},
+	} {
+		test.request.Kind = promptkit.SmokeTestTextInference
+		report, err := engine.SmokeTest(context.Background(), test.request)
+		if err != nil || report.Target.ProfileID != test.profile || report.Target.EffectiveSettings.Model != test.model {
+			t.Fatalf("report = %+v, %v", report, err)
+		}
+		if strings.Contains(client.requests[len(client.requests)-1].Prompt.Messages[0].Content, "private application marker") {
+			t.Fatal("application content used by diagnostic")
+		}
+	}
+}

capacity_contract_test.go

@@ -7,7 +7,8 @@ import (
 	"testing"
 	"time"
 
-	"gitea.maximumdirect.net/eric/promptkit"
+	"go.maximumdirect.net/configkit"
+	"go.maximumdirect.net/promptkit"
 )
 
 func TestEngineLimitsInjectedClientConcurrency(t *testing.T) {
@@ -150,17 +151,7 @@ func TestUnlimitedBackendsRetainInjectedClientConcurrency(t *testing.T) {
 		{
 			name: "endpoint-only profile",
 			configure: func(t *testing.T, client promptkit.LLMClient) *promptkit.Engine {
-				engine, err := promptkit.NewEngine(promptkit.Config{},
-					promptkit.WithPromptFS(contractPromptFS("prompt", "profile", "message"), "."),
-					promptkit.WithProfiles(promptkit.Profile{
-						ID: "profile", Endpoint: "http://endpoint.example/v1", Model: "model",
-					}),
-					promptkit.WithLLMClient(client),
-				)
-				if err != nil {
-					t.Fatalf("construct endpoint-only engine: %v", err)
-				}
-				return engine
+				return newSimpleEngine(t, "message", promptkit.WithLLMClient(client))
 			},
 		},
 	}
@@ -250,27 +241,36 @@ func newBackendCapacityEngine(
 	reader promptkit.ArtifactReader,
 ) *promptkit.Engine {
 	t.Helper()
-	promptFS := contractPromptFS("prompt", "profile", "message")
+	promptID, content := "prompt", "message"
+	var inputs []any
 	if reader != nil {
-		promptFS = contractInputPromptFS()
+		promptID, content = "input-prompt", `{{input "input"}}`
+		inputs = []any{map[string]any{"name": "input", "required": true}}
 	}
-	options := []promptkit.Option{
-		promptkit.WithPromptFS(promptFS, "."),
-		promptkit.WithBackend(promptkit.Backend{
-			ID:               "limited",
-			Endpoint:         "http://backend.example/v1",
-			ConcurrencyLimit: limit,
-			QueueCapacity:    queueCapacity,
-		}),
-		promptkit.WithProfiles(promptkit.Profile{
-			ID: "profile", BackendID: "limited", Model: "model",
-		}),
-		promptkit.WithLLMClient(client),
+	backend := map[string]any{"id": "limited", "endpoint": "http://backend.example/v1", "concurrency_limit": limit}
+	if queueCapacity != nil {
+		backend["queue_capacity"] = *queueCapacity
 	}
+	definition := map[string]any{
+		"id":              promptID,
+		"default_profile": "profile",
+		"messages":        []any{map[string]any{"role": "user", "content": content}},
+		"output":          map[string]any{"format": "text", "validation_mode": "none"},
+	}
+	if inputs != nil {
+		definition["inputs"] = inputs
+	}
+	options := []promptkit.Option{promptkit.WithLLMClient(client)}
 	if reader != nil {
 		options = append(options, promptkit.WithArtifactReader(reader))
 	}
-	engine, err := promptkit.NewEngine(promptkit.Config{}, options...)
+	engine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{Main: &configkit.Input{
+		Source: configkit.Mapping("capacity fixture", map[string]any{
+			"backends": []any{backend},
+			"profiles": []any{map[string]any{"id": "profile", "backend": "limited", "model": "model"}},
+			"prompts":  []any{definition},
+		}),
+	}}}, options...)
 	if err != nil {
 		t.Fatalf("construct capacity engine: %v", err)
 	}

configuration_diff_contract_test.go

@@ -0,0 +1,261 @@
+package promptkit_test
+
+import (
+	"errors"
+	"strings"
+	"testing"
+
+	"go.maximumdirect.net/configkit"
+	"go.maximumdirect.net/promptkit"
+)
+
+func TestConfigurationInspectionDiffsCapturedValues(t *testing.T) {
+	left := comparisonInspection(t, "earlier-model")
+	right := comparisonInspection(t, "later-model")
+
+	differences, err := left.Diff(right)
+	if err != nil {
+		t.Fatalf("compare captured configurations: %v", err)
+	}
+	if len(differences) != 1 {
+		t.Fatalf("difference count = %d, want 1: %+v", len(differences), differences)
+	}
+	difference := differences[0]
+	if got, want := difference.Path.String(), "/profiles"; got != want {
+		t.Fatalf("difference path = %q, want %q", got, want)
+	}
+	if !strings.Contains(difference.Left.YAML, "earlier-model") || !strings.Contains(difference.Right.YAML, "later-model") {
+		t.Fatalf("difference snapshots = (%q, %q)", difference.Left.YAML, difference.Right.YAML)
+	}
+
+	reversed, err := right.Diff(left)
+	if err != nil {
+		t.Fatalf("compare reversed configurations: %v", err)
+	}
+	if len(reversed) != 1 || reversed[0].Path.String() != difference.Path.String() ||
+		!strings.Contains(reversed[0].Left.YAML, "later-model") ||
+		!strings.Contains(reversed[0].Right.YAML, "earlier-model") {
+		t.Fatalf("reversed difference = %+v", reversed)
+	}
+
+	equal := comparisonInspection(t, "earlier-model")
+	if equal, err := left.Diff(equal); err != nil || len(equal) != 0 {
+		t.Fatalf("compare equal configurations = (%+v, %v), want no differences", equal, err)
+	}
+	if equal, err := left.Diff(left); err != nil || len(equal) != 0 {
+		t.Fatalf("compare view with itself = (%+v, %v), want no differences", equal, err)
+	}
+}
+
+func TestConfigurationInspectionDiffRejectsZeroViews(t *testing.T) {
+	valid := comparisonInspection(t, "current-model")
+	var zero promptkit.ConfigurationInspection
+
+	for _, comparison := range []struct {
+		name        string
+		left, right promptkit.ConfigurationInspection
+	}{
+		{name: "zero left", left: zero, right: valid},
+		{name: "zero right", left: valid, right: zero},
+		{name: "both zero", left: zero, right: zero},
+	} {
+		t.Run(comparison.name, func(t *testing.T) {
+			differences, err := comparison.left.Diff(comparison.right)
+			if !errors.Is(err, promptkit.ErrInvalidConfig) || len(differences) != 0 {
+				t.Fatalf("Diff() = (%+v, %v), want ErrInvalidConfig and no records", differences, err)
+			}
+		})
+	}
+}
+
+func TestConfigurationInspectionDiffIgnoresSourceHistory(t *testing.T) {
+	left := comparisonInspectionFrom(t, "earlier source", "same-model", nil)
+	right := comparisonInspectionFrom(t, "later source", "same-model", nil)
+	if !hasSourceLabel(left, "earlier source") || !hasSourceLabel(right, "later source") {
+		t.Fatal("comparison fixtures did not retain their distinct source labels")
+	}
+
+	differences, err := left.Diff(right)
+	if err != nil || len(differences) != 0 {
+		t.Fatalf("compare equal effective values with different sources = (%+v, %v), want no differences", differences, err)
+	}
+}
+
+func TestConfigurationInspectionDiffUsesNormalizedValues(t *testing.T) {
+	inspect := func(endpoint string) promptkit.ConfigurationInspection {
+		t.Helper()
+		engine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{Underlays: []configkit.Underlay{{Inputs: []configkit.Input{{Source: configkit.Mapping("backend", map[string]any{
+			"backends": []any{map[string]any{"id": "local", "endpoint": endpoint}},
+		})}}}}}})
+		if err != nil {
+			t.Fatal(err)
+		}
+		view, err := engine.InspectConfiguration()
+		if err != nil {
+			t.Fatal(err)
+		}
+		return view
+	}
+	left := inspect(" HTTPS://example.invalid/v1 ")
+	equal := inspect("https://example.invalid/v1")
+	differences, err := left.Diff(equal)
+	if err != nil || len(differences) != 0 {
+		t.Fatalf("normalized diff = %+v, %v", differences, err)
+	}
+	changed := inspect("https://other.invalid/v1")
+	differences, err = left.Diff(changed)
+	if err != nil || len(differences) != 1 || differences[0].Path.String() != "/backends" {
+		t.Fatalf("changed diff = %+v, %v", differences, err)
+	}
+}
+
+func TestConfigurationInspectionDiffReportsPresenceAndSequences(t *testing.T) {
+	left := comparisonInspectionFrom(t, "before", "same-model", map[string]any{
+		"sequence": []any{"earlier", "unchanged"},
+	})
+	right := comparisonInspectionFrom(t, "after", "same-model", map[string]any{
+		"nullable": nil,
+		"sequence": []any{"later", "unchanged"},
+	})
+
+	differences, err := left.Diff(right)
+	if err != nil {
+		t.Fatalf("compare presence and sequence values: %v", err)
+	}
+	if len(differences) != 1 || differences[0].Path.String() != "/backends" {
+		t.Fatalf("atomic backend list difference = %+v", differences)
+	}
+	difference := differences[0]
+	if !difference.Left.Present || !difference.Right.Present ||
+		strings.Contains(difference.Left.YAML, "nullable:") || !strings.Contains(difference.Right.YAML, "nullable: null") ||
+		!strings.Contains(difference.Left.YAML, "earlier") || !strings.Contains(difference.Right.YAML, "later") {
+		t.Fatalf("atomic sequence snapshots = %+v", difference)
+	}
+}
+
+func TestConfigurationInspectionDiffProtectsSensitiveValuesAndReturnedRecords(t *testing.T) {
+	left := comparisonInspectionFrom(t, "before", "same-model", map[string]any{
+		"token":   "left-secret-sentinel",
+		"visible": "before-value",
+	})
+	right := comparisonInspectionFrom(t, "after", "same-model", map[string]any{
+		"token":   "right-secret-sentinel",
+		"visible": "after-value",
+	})
+
+	differences, err := left.Diff(right)
+	if err != nil {
+		t.Fatalf("compare sensitive values: %v", err)
+	}
+	if len(differences) != 1 || differences[0].Path.String() != "/backends" {
+		t.Fatalf("sensitive backend list difference = %+v", differences)
+	}
+	secret := differences[0]
+	if !secret.Left.Redacted || !secret.Right.Redacted || !secret.HiddenChanged {
+		t.Fatalf("sensitive difference = %+v", secret)
+	}
+	if strings.Contains(secret.Left.YAML, "left-secret-sentinel") || strings.Contains(secret.Left.YAML, "right-secret-sentinel") ||
+		strings.Contains(secret.Right.YAML, "left-secret-sentinel") || strings.Contains(secret.Right.YAML, "right-secret-sentinel") {
+		t.Fatalf("sensitive snapshots disclosed a value: %+v", secret)
+	}
+	if !strings.Contains(secret.Left.YAML, "before-value") || !strings.Contains(secret.Right.YAML, "after-value") {
+		t.Fatalf("public sibling missing from snapshots: %+v", secret)
+	}
+
+	differences[0].Path = configkit.Keys("caller", "mutation")
+	differences[0].Left.YAML = "caller mutation"
+	differences[0].Right.YAML = "caller mutation"
+	again, err := left.Diff(right)
+	if err != nil {
+		t.Fatalf("repeat comparison after mutating returned records: %v", err)
+	}
+	if len(again) != 1 {
+		t.Fatalf("repeat difference count = %d, want 1: %+v", len(again), again)
+	}
+	againByPath := differencesByPath(again)
+	if _, exists := againByPath["/caller/mutation"]; exists {
+		t.Fatal("mutating returned path reached retained inspection")
+	}
+	if _, exists := againByPath["/backends"]; !exists {
+		t.Fatalf("original backend path was lost after mutation: %+v", again)
+	}
+	for _, difference := range again {
+		if difference.Left.YAML == "caller mutation" || difference.Right.YAML == "caller mutation" {
+			t.Fatalf("mutating returned snapshots reached retained inspection: %+v", difference)
+		}
+	}
+}
+
+func TestConfigurationInspectionDiffMasksSensitiveShapeReplacement(t *testing.T) {
+	left := comparisonInspectionFrom(t, "before", "same-model", map[string]any{
+		"details": map[string]any{
+			"token": "nested-secret-sentinel",
+			"label": "public detail",
+		},
+	})
+	right := comparisonInspectionFrom(t, "after", "same-model", map[string]any{
+		"details": "replacement",
+	})
+
+	differences, err := left.Diff(right)
+	if err != nil {
+		t.Fatalf("compare sensitive shape replacement: %v", err)
+	}
+	if len(differences) != 1 || differences[0].Path.String() != "/backends" {
+		t.Fatalf("shape replacement differences = %+v", differences)
+	}
+	difference := differences[0]
+	if !difference.Left.Redacted || !difference.Right.Redacted || !difference.HiddenChanged {
+		t.Fatalf("shape replacement masking = %+v", difference)
+	}
+	if strings.Contains(difference.Left.YAML, "nested-secret-sentinel") || strings.Contains(difference.Right.YAML, "nested-secret-sentinel") {
+		t.Fatalf("shape replacement disclosed a sensitive value: %+v", difference)
+	}
+}
+
+func comparisonInspection(t *testing.T, model string) promptkit.ConfigurationInspection {
+	t.Helper()
+	return comparisonInspectionFrom(t, "comparison fixture", model, nil)
+}
+
+func comparisonInspectionFrom(t *testing.T, source, model string, extraParams map[string]any) promptkit.ConfigurationInspection {
+	t.Helper()
+	backend := map[string]any{"id": "local", "endpoint": "https://example.invalid/v1"}
+	if extraParams != nil {
+		backend["extra_params"] = extraParams
+	}
+	engine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{
+		Underlays: []configkit.Underlay{{Inputs: []configkit.Input{{Source: configkit.Mapping(source, map[string]any{
+			"backends": []any{backend},
+			"profiles": []any{map[string]any{
+				"id": "offline", "backend": "local",
+				"model": model,
+			}},
+		})}}}},
+	}})
+	if err != nil {
+		t.Fatalf("construct comparison engine: %v", err)
+	}
+	inspection, err := engine.InspectConfiguration()
+	if err != nil {
+		t.Fatalf("inspect comparison engine: %v", err)
+	}
+	return inspection
+}
+
+func differencesByPath(differences []configkit.Difference) map[string]configkit.Difference {
+	byPath := make(map[string]configkit.Difference, len(differences))
+	for _, difference := range differences {
+		byPath[difference.Path.String()] = difference
+	}
+	return byPath
+}
+
+func hasSourceLabel(view promptkit.ConfigurationInspection, label string) bool {
+	for _, source := range view.Sources() {
+		if source.Label == label {
+			return true
+		}
+	}
+	return false
+}

configuration_inspection.go

@@ -0,0 +1,99 @@
+package promptkit
+
+import (
+	"fmt"
+	"io"
+
+	"go.maximumdirect.net/configkit"
+	"go.maximumdirect.net/promptkit/internal/configuration"
+)
+
+// ConfigurationInspection is a read-only view of one Engine's captured
+// effective configuration and source history. It does not expose typed
+// configuration values, filesystem contexts, or runtime credentials.
+// Report content may include application prompt and provider data, so callers
+// choose where and to whom to disclose it.
+type ConfigurationInspection struct {
+	source configuration.Inspection
+}
+
+// InspectConfiguration returns the ConfigKit inspection view captured by this
+// Engine. It does not reload sources, resolve credentials, or contact a model.
+// A nil Engine returns ErrInvalidConfig.
+func (e *Engine) InspectConfiguration() (ConfigurationInspection, error) {
+	if e == nil || e.internal == nil {
+		return ConfigurationInspection{}, fmt.Errorf("%w: engine is nil", ErrInvalidConfig)
+	}
+	return ConfigurationInspection{source: e.internal.InspectConfiguration()}, nil
+}
+
+// Diff compares this view's captured effective configuration with other. This
+// view is the earlier (left) value and other is the later (right) value. The
+// comparison delegates to ConfigKit's frozen effective-value semantics, so
+// source or resolution history alone does not create differences. It does not
+// reread external sources or compare referenced file contents, resolved
+// credential contents, later execution preparation, or request-specific
+// overrides. Either zero view returns ErrInvalidConfig and no records; other
+// ConfigKit errors are returned unchanged. An empty diff is not evidence of
+// equivalent engine behavior or execution. Records can contain ordinary
+// application data, so callers choose where and to whom to disclose them.
+func (v ConfigurationInspection) Diff(other ConfigurationInspection) ([]configkit.Difference, error) {
+	if v.source == nil || other.source == nil {
+		return nil, ErrInvalidConfig
+	}
+	return configuration.Diff(v.source, other.source)
+}
+
+// Metadata returns detached effective presence and history records.
+func (v ConfigurationInspection) Metadata() configkit.Metadata {
+	if v.source == nil {
+		return configkit.Metadata{}
+	}
+	return v.source.Metadata()
+}
+
+// Sources returns a detached list of loaded source descriptions.
+func (v ConfigurationInspection) Sources() []configkit.SourceInfo {
+	if v.source == nil {
+		return nil
+	}
+	return v.source.Sources()
+}
+
+// Diagnostics returns detached nonfatal load diagnostics.
+func (v ConfigurationInspection) Diagnostics() []configkit.Diagnostic {
+	if v.source == nil {
+		return nil
+	}
+	return v.source.Diagnostics()
+}
+
+// Explain returns detached effective and historical values for a known path.
+// Dynamic credential-shaped provider keys are redacted by ConfigKit; ordinary
+// prompt and provider content is not automatically safe to disclose.
+func (v ConfigurationInspection) Explain(path configkit.Path) (configkit.Explanation, bool) {
+	if v.source == nil {
+		return configkit.Explanation{}, false
+	}
+	return v.source.Explain(path)
+}
+
+// WriteEffective writes ConfigKit's redacted effective YAML. Writer errors
+// remain available through errors.Is and errors.As. The zero view returns
+// ErrInvalidConfig.
+func (v ConfigurationInspection) WriteEffective(w io.Writer) error {
+	if v.source == nil {
+		return ErrInvalidConfig
+	}
+	return v.source.WriteEffective(w)
+}
+
+// WriteExplanation writes ConfigKit's redacted history for path. Unknown paths
+// produce no output; writer errors remain inspectable. The zero view returns
+// ErrInvalidConfig.
+func (v ConfigurationInspection) WriteExplanation(w io.Writer, path configkit.Path) error {
+	if v.source == nil {
+		return ErrInvalidConfig
+	}
+	return v.source.WriteExplanation(w, path)
+}

configuration_inspection_contract_test.go

@@ -0,0 +1,321 @@
+package promptkit_test
+
+import (
+	"bytes"
+	"context"
+	"errors"
+	"fmt"
+	"io"
+	"strings"
+	"sync"
+	"sync/atomic"
+	"testing"
+	"testing/fstest"
+
+	"go.maximumdirect.net/configkit"
+	"go.maximumdirect.net/promptkit"
+)
+
+func TestConfigurationInspectionReportsLiteralKeysAndRedactedHistory(t *testing.T) {
+	engine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{
+		Underlays: []configkit.Underlay{
+			{Inputs: []configkit.Input{{Source: configkit.Mapping("defaults", map[string]any{
+				"backends": []any{map[string]any{
+					"id": "local", "endpoint": " HTTPS://example.invalid/v1 ", "api_key_env": "OPTIONAL_KEY",
+					"extra_params": map[string]any{"token": "private-sentinel", "visible": "public-value"},
+				}},
+				"collections": map[string]any{"literal": map[string]any{
+					"backends": []any{
+						map[string]any{"id": "local", "endpoint": "https://example.invalid/v1", "extra_params": map[string]any{"token": "namespaced-private-sentinel"}},
+						map[string]any{"id": "child", "extends": map[string]any{"id": "local"}},
+					},
+					"profiles": []any{
+						map[string]any{"id": "base.v1", "backend": "local", "model": "default-model"},
+						map[string]any{"id": "id", "extends": map[string]any{"id": "base.v1"}},
+					},
+				}},
+			})}}},
+			{Inputs: []configkit.Input{{Source: configkit.Mapping("operator", map[string]any{
+				"collections": map[string]any{"literal": map[string]any{
+					"backends": []any{map[string]any{"id": "local", "endpoint": "https://example.invalid/v1", "extra_params": map[string]any{"token": "replacement-private-sentinel"}}},
+					"profiles": []any{map[string]any{"id": "base.v1", "backend": "local", "model": "operator-model"}},
+				}},
+			})}}},
+		},
+	}}, promptkit.WithBundledOpenRouter(false))
+	if err != nil {
+		t.Fatalf("construct layered engine: %v", err)
+	}
+	view, err := engine.InspectConfiguration()
+	if err != nil {
+		t.Fatalf("inspect configuration: %v", err)
+	}
+	modelPath := configkit.Keys("collections", "literal", "profiles").Index(0).Key("model")
+	model, found := view.Explain(modelPath)
+	if !found || !strings.Contains(model.Effective.YAML, "operator-model") || len(model.History) == 0 {
+		t.Fatalf("layered model explanation = (%+v, %t)", model, found)
+	}
+	inheritedPath := configkit.Keys("collections", "literal", "profiles").Index(1).Key("model")
+	inherited, found := view.Explain(inheritedPath)
+	if !found || !strings.Contains(inherited.Effective.YAML, "operator-model") || len(inherited.History) == 0 || !view.Metadata().IsExplicit(inheritedPath) {
+		t.Fatalf("literal-key inheritance explanation = (%+v, %t)", inherited, found)
+	}
+	if inheritedPath.String() != "/collections/literal/profiles/1/model" {
+		t.Fatalf("literal-key path = %q", inheritedPath)
+	}
+	secretPath := configkit.Keys("backends").Index(0).Key("extra_params").Key("token")
+	endpointPath := configkit.Keys("backends").Index(0).Key("endpoint")
+	endpoint, found := view.Explain(endpointPath)
+	if !found || !strings.Contains(endpoint.Effective.YAML, "https://example.invalid/v1") || strings.Contains(fmt.Sprintf("%+v", view.Metadata().History(endpointPath)), "private-sentinel") {
+		t.Fatalf("normalized endpoint explanation = %+v", endpoint)
+	}
+	secret, found := view.Explain(secretPath)
+	if !found || !secret.Effective.Redacted || strings.Contains(secret.Effective.YAML, "private-sentinel") {
+		t.Fatalf("credential-shaped provider value was not redacted: (%+v, %t)", secret, found)
+	}
+	namespacedSecretPath := configkit.Keys("collections", "literal", "backends").Index(0).Key("extra_params").Key("token")
+	namespacedSecret, found := view.Explain(namespacedSecretPath)
+	if !found || !namespacedSecret.Effective.Redacted || strings.Contains(namespacedSecret.Effective.YAML, "namespaced-private-sentinel") {
+		t.Fatalf("namespaced provider value was not redacted: (%+v, %t)", namespacedSecret, found)
+	}
+	inheritedSecretPath := configkit.Keys("collections", "literal", "backends").Index(1).Key("extra_params").Key("token")
+	inheritedSecret, found := view.Explain(inheritedSecretPath)
+	if !found || !inheritedSecret.Effective.Redacted || strings.Contains(inheritedSecret.Effective.YAML, "replacement-private-sentinel") {
+		t.Fatalf("inherited provider value was not redacted: (%+v, %t)", inheritedSecret, found)
+	}
+	metadata := fmt.Sprintf("%+v", view.Metadata().History(inheritedSecretPath))
+	if strings.Contains(metadata, "replacement-private-sentinel") || strings.Contains(metadata, "namespaced-private-sentinel") {
+		t.Fatal("metadata disclosed a sensitive provider value")
+	}
+	var effective bytes.Buffer
+	if err := view.WriteEffective(&effective); err != nil {
+		t.Fatalf("write effective configuration: %v", err)
+	}
+	if strings.Contains(effective.String(), "private-sentinel") || !strings.Contains(effective.String(), "OPTIONAL_KEY") || !strings.Contains(effective.String(), "public-value") {
+		t.Fatalf("unexpected effective report: %s", effective.String())
+	}
+	var explanation bytes.Buffer
+	if err := view.WriteExplanation(&explanation, namespacedSecretPath); err != nil || strings.Contains(explanation.String(), "namespaced-private-sentinel") {
+		t.Fatalf("secret explanation = (%q, %v)", explanation.String(), err)
+	}
+
+	sources := view.Sources()
+	if len(sources) == 0 {
+		t.Fatal("inspection returned no sources")
+	}
+	originalLabel := sources[0].Label
+	sources[0].Label = "changed"
+	if view.Sources()[0].Label != originalLabel {
+		t.Fatal("source mutation reached retained inspection")
+	}
+	model.History[0].After.YAML = "changed"
+	if again, ok := view.Explain(modelPath); !ok || again.History[0].After.YAML == "changed" {
+		t.Fatal("explanation mutation reached retained inspection")
+	}
+	history := view.Metadata().History(modelPath)
+	if len(history) == 0 {
+		t.Fatal("metadata history is empty")
+	}
+	history[0].Reasons = []string{"changed"}
+	if again := view.Metadata().History(modelPath); len(again) == 0 || (len(again[0].Reasons) > 0 && again[0].Reasons[0] == "changed") {
+		t.Fatal("metadata mutation reached retained inspection")
+	}
+}
+
+func TestConfigurationInspectionFailureDoesNotDiscloseSensitiveValues(t *testing.T) {
+	_, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{Main: &configkit.Input{
+		Source: configkit.Mapping("invalid backend", map[string]any{
+			"collections": map[string]any{"private": map[string]any{"backends": []any{
+				map[string]any{"id": "backend", "endpoint": "invalid endpoint", "extra_params": map[string]any{"token": "failure-private-sentinel"}},
+			}}},
+		}),
+	}}})
+	var loadErr *configkit.LoadError
+	if !errors.As(err, &loadErr) {
+		t.Fatalf("invalid backend error = %v, want load error", err)
+	}
+	if text := fmt.Sprintf("%v %+v", err, loadErr.Diagnostics()); strings.Contains(text, "failure-private-sentinel") {
+		t.Fatalf("failure diagnostics disclosed a sensitive value: %s", text)
+	}
+}
+
+func TestConfigurationInspectionRetainsMainSpecificationDeclaration(t *testing.T) {
+	engine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{Main: &configkit.Input{
+		Source: configkit.Mapping("main", map[string]any{"version": "2"}),
+	}}})
+	if err != nil {
+		t.Fatalf("construct engine with accepted declaration: %v", err)
+	}
+	inspection, err := engine.InspectConfiguration()
+	if err != nil {
+		t.Fatalf("inspect configuration: %v", err)
+	}
+	path := configkit.Keys("version")
+	field, found := inspection.Explain(path)
+	if !found || !inspection.Metadata().IsExplicit(path) || !strings.Contains(field.Effective.YAML, "2") {
+		t.Fatalf("main specification declaration was not retained: (%+v, %t)", field, found)
+	}
+	var effective bytes.Buffer
+	if err := inspection.WriteEffective(&effective); err != nil || !strings.Contains(effective.String(), "version") {
+		t.Fatalf("effective inspection = (%q, %v), want retained version", effective.String(), err)
+	}
+}
+
+type failingInspectionWriter struct{ err error }
+
+func (writer failingInspectionWriter) Write([]byte) (int, error) { return 0, writer.err }
+
+func TestConfigurationInspectionZeroAndWriterErrors(t *testing.T) {
+	var engine *promptkit.Engine
+	if _, err := engine.InspectConfiguration(); !errors.Is(err, promptkit.ErrInvalidConfig) {
+		t.Fatalf("nil engine inspection = %v", err)
+	}
+	var empty promptkit.ConfigurationInspection
+	path := configkit.Keys("profiles", "missing")
+	if _, found := empty.Metadata().Lookup(path); found || len(empty.Sources()) != 0 || len(empty.Diagnostics()) != 0 {
+		t.Fatal("zero inspection reported configuration")
+	}
+	if _, found := empty.Explain(path); found {
+		t.Fatal("zero inspection explained an unknown path")
+	}
+	if err := empty.WriteEffective(io.Discard); !errors.Is(err, promptkit.ErrInvalidConfig) {
+		t.Fatalf("zero effective writer = %v", err)
+	}
+	if err := empty.WriteExplanation(io.Discard, path); !errors.Is(err, promptkit.ErrInvalidConfig) {
+		t.Fatalf("zero explanation writer = %v", err)
+	}
+
+	profile := configkit.Input{Source: configkit.Mapping("inspection test profile", map[string]any{
+		"profiles": []any{map[string]any{"id": "inspection-profile", "endpoint": "https://configured.example/v1", "model": "configured-model"}},
+	})}
+	valid, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{
+		Underlays: []configkit.Underlay{{Inputs: []configkit.Input{profile}}},
+	}})
+	if err != nil {
+		t.Fatalf("construct engine: %v", err)
+	}
+	view, err := valid.InspectConfiguration()
+	if err != nil {
+		t.Fatalf("inspect configuration: %v", err)
+	}
+	writeErr := errors.New("report writer failed")
+	writer := failingInspectionWriter{err: writeErr}
+	if err := view.WriteEffective(writer); !errors.Is(err, writeErr) {
+		t.Fatalf("effective writer error = %v", err)
+	}
+	if err := view.WriteExplanation(writer, configkit.Keys("profiles").Index(0).Key("model")); !errors.Is(err, writeErr) {
+		t.Fatalf("explanation writer error = %v", err)
+	}
+}
+
+func TestConfigurationInspectionUsesCapturedSourcesWithoutProviderCalls(t *testing.T) {
+	const promptDocument = "prompts:\n  - id: captured\n    default_profile: offline\n    messages:\n      - role: user\n        content_file: body.tmpl\n    output:\n      format: text\n      validation_mode: none\n"
+	source := &unreadableAfterConstructionFS{FS: fstest.MapFS{
+		"config/prompt.yaml": &fstest.MapFile{Data: []byte(promptDocument)},
+		"config/body.tmpl":   &fstest.MapFile{Data: []byte("hello")},
+	}}
+	inputs, err := promptkit.FilesystemInputs("inspection source", source, "config")
+	if err != nil {
+		t.Fatalf("select inputs: %v", err)
+	}
+	otherSource := &unreadableAfterConstructionFS{FS: fstest.MapFS{
+		"config/prompt.yaml": &fstest.MapFile{Data: []byte(promptDocument)},
+		"config/body.tmpl":   &fstest.MapFile{Data: []byte("hello")},
+	}}
+	otherInputs, err := promptkit.FilesystemInputs("comparison source", otherSource, "config")
+	if err != nil {
+		t.Fatalf("select comparison inputs: %v", err)
+	}
+	var modelCalls atomic.Int64
+	client := countingLLMClient{calls: &modelCalls}
+	endpoint, model, otherModel := "https://example.invalid/v1", "offline-model", "comparison-model"
+	engine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{
+		Underlays: []configkit.Underlay{{Inputs: inputs}, {Inputs: []configkit.Input{{Source: configkit.Mapping("offline profile", map[string]any{
+			"profiles": []any{map[string]any{"id": "offline", "endpoint": endpoint, "model": model}},
+		})}}}},
+	}}, promptkit.WithLLMClient(client))
+	if err != nil {
+		t.Fatalf("construct engine: %v", err)
+	}
+	otherEngine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{
+		Underlays: []configkit.Underlay{{Inputs: otherInputs}, {Inputs: []configkit.Input{{Source: configkit.Mapping("comparison profile", map[string]any{
+			"profiles": []any{map[string]any{"id": "offline", "endpoint": endpoint, "model": otherModel}},
+		})}}}},
+	}}, promptkit.WithLLMClient(client))
+	if err != nil {
+		t.Fatalf("construct comparison engine: %v", err)
+	}
+	source.closed.Store(true)
+	otherSource.closed.Store(true)
+	view, err := engine.InspectConfiguration()
+	if err != nil {
+		t.Fatalf("inspect configuration: %v", err)
+	}
+	otherView, err := otherEngine.InspectConfiguration()
+	if err != nil {
+		t.Fatalf("inspect comparison configuration: %v", err)
+	}
+	var expected configkit.Difference
+	for index := 0; index < 3; index++ {
+		differences, err := view.Diff(otherView)
+		if err != nil || len(differences) != 1 {
+			t.Fatalf("compare captured configurations = (%+v, %v), want one difference", differences, err)
+		}
+		if differences[0].Path.String() != "/profiles" ||
+			!strings.Contains(differences[0].Left.YAML, model) ||
+			!strings.Contains(differences[0].Right.YAML, otherModel) {
+			t.Fatalf("captured difference = %+v", differences[0])
+		}
+		if index == 0 {
+			expected = differences[0]
+		} else if differences[0].Path.String() != expected.Path.String() ||
+			differences[0].Left.YAML != expected.Left.YAML || differences[0].Right.YAML != expected.Right.YAML {
+			t.Fatalf("captured difference changed between reads: first=%+v next=%+v", expected, differences[0])
+		}
+	}
+	if _, found := view.Explain(configkit.Keys("prompts").Index(0).Key("messages").Index(0).Key("content_file")); !found {
+		t.Fatal("captured prompt content path was not inspectable")
+	}
+	var report bytes.Buffer
+	if err := view.WriteEffective(&report); err != nil || !strings.Contains(report.String(), "captured") {
+		t.Fatalf("write captured report = (%q, %v)", report.String(), err)
+	}
+	if source.reads.Load() != 0 || otherSource.reads.Load() != 0 || modelCalls.Load() != 0 {
+		t.Fatalf("inspection or comparison touched source or provider: reads=(%d,%d) requests=%d", source.reads.Load(), otherSource.reads.Load(), modelCalls.Load())
+	}
+
+	const calls = 24
+	var wait sync.WaitGroup
+	errs := make(chan error, calls)
+	for index := 0; index < calls; index++ {
+		wait.Add(1)
+		go func(index int) {
+			defer wait.Done()
+			switch index % 3 {
+			case 0:
+				_, err := engine.InspectConfiguration()
+				errs <- err
+			case 1:
+				_, err := engine.Run(context.Background(), promptkit.RunRequest{PromptID: "captured", SessionID: "direct"})
+				errs <- err
+			case 2:
+				differences, err := view.Diff(otherView)
+				if err == nil && (len(differences) != 1 || differences[0].Path.String() != expected.Path.String() ||
+					differences[0].Left.YAML != expected.Left.YAML || differences[0].Right.YAML != expected.Right.YAML) {
+					err = errors.New("concurrent comparison returned an unexpected difference")
+				}
+				errs <- err
+			}
+		}(index)
+	}
+	wait.Wait()
+	close(errs)
+	for err := range errs {
+		if err != nil {
+			t.Fatal(err)
+		}
+	}
+	if source.reads.Load() != 0 || otherSource.reads.Load() != 0 || modelCalls.Load() != calls/3 {
+		t.Fatalf("concurrent inspection/comparison/execution = source reads (%d,%d), model calls %d", source.reads.Load(), otherSource.reads.Load(), modelCalls.Load())
+	}
+}

configuration_sources.go

@@ -0,0 +1,100 @@
+package promptkit
+
+import (
+	"context"
+	"fmt"
+	"io/fs"
+	"os"
+	"path/filepath"
+
+	"go.maximumdirect.net/configkit"
+	"go.maximumdirect.net/promptkit/internal/domain"
+	"go.maximumdirect.net/promptkit/internal/filecatalog"
+)
+
+// CollectionInput mounts source at the selected collection's configuration
+// destination, collections/<namespace>, and enables ConfigKit imports for that
+// input. An invalid namespace returns an error matching ErrInvalidConfig. The
+// source is not read or validated until ConfigKit loads it. Callers may adjust
+// the returned input, including its filesystem context or imports setting.
+func CollectionInput(namespace string, source configkit.Source) (configkit.Input, error) {
+	if !domain.IsValidResourceNamespace(namespace) {
+		return configkit.Input{}, fmt.Errorf("%w: invalid collection namespace %q", ErrInvalidConfig, namespace)
+	}
+	return configkit.Input{
+		Source:      source,
+		Destination: configkit.Keys("collections", namespace),
+		Imports:     true,
+	}, nil
+}
+
+// DirectoryInputs selects the YAML files beneath an explicitly chosen host
+// directory as one strict-composition group. The returned order is deterministic;
+// it does not assign precedence among those files. Discovery is recursive over
+// .yaml and .yml files. An unavailable root fails; an empty directory yields
+// no inputs. Imports are disabled unless callers enable them on the inputs.
+func DirectoryInputs(directory string) ([]configkit.Input, error) {
+	root, err := filepath.Abs(directory)
+	if err != nil {
+		return nil, fmt.Errorf("resolve configuration directory: %w", err)
+	}
+	info, err := os.Stat(root)
+	if err != nil {
+		return nil, fmt.Errorf("inspect configuration directory: %w", err)
+	}
+	if !info.IsDir() {
+		return nil, fmt.Errorf("configuration root is not a directory: %s", root)
+	}
+	root, err = filepath.EvalSymlinks(root)
+	if err != nil {
+		return nil, fmt.Errorf("resolve configuration directory: %w", err)
+	}
+	files, err := filecatalog.FindYAMLFiles(context.Background(), root)
+	if err != nil {
+		return nil, fmt.Errorf("discover configuration files: %w", err)
+	}
+	inputs := make([]configkit.Input, 0, len(files))
+	for _, file := range files {
+		inputs = append(inputs, configkit.Input{
+			Source:     configkit.File(file),
+			Filesystem: &configkit.FilesystemContext{Root: root},
+		})
+	}
+	return inputs, nil
+}
+
+// FilesystemInputs selects the YAML files beneath root in a caller-owned
+// filesystem as one strict-composition group. Discovery is recursive over
+// .yaml and .yml files, with deterministic ordering and no precedence among
+// files. The caller retains ownership of the filesystem; an unavailable root
+// fails and an empty directory yields no inputs. Imports remain disabled unless
+// callers enable them on inputs.
+func FilesystemInputs(label string, filesystem fs.FS, root string) ([]configkit.Input, error) {
+	if filesystem == nil {
+		return nil, fmt.Errorf("configuration filesystem is nil")
+	}
+	root = filecatalog.CleanFSRoot(root)
+	if !fs.ValidPath(root) {
+		return nil, fmt.Errorf("configuration root is not a valid filesystem path: %s", root)
+	}
+	info, err := fs.Stat(filesystem, root)
+	if err != nil {
+		return nil, fmt.Errorf("inspect configuration root: %w", err)
+	}
+	if !info.IsDir() {
+		return nil, fmt.Errorf("configuration root is not a directory: %s", root)
+	}
+	files, err := filecatalog.FindFSYAMLFiles(context.Background(), filesystem, root)
+	if err != nil {
+		return nil, fmt.Errorf("discover configuration files: %w", err)
+	}
+	namespace := configkit.NewFilesystem(label, filesystem)
+	inputs := make([]configkit.Input, 0, len(files))
+	for _, file := range files {
+		inputs = append(inputs, configkit.Input{
+			Source:     namespace.File(file),
+			Filesystem: &configkit.FilesystemContext{FS: &namespace, Root: root},
+		})
+	}
+	return inputs, nil
+}

configuration_sources_test.go

@@ -0,0 +1,157 @@
+package promptkit_test
+
+import (
+	"context"
+	"errors"
+	"io/fs"
+	"os"
+	"path/filepath"
+	"testing"
+	"testing/fstest"
+
+	"go.maximumdirect.net/configkit"
+	"go.maximumdirect.net/promptkit"
+)
+
+type sourceDocument struct {
+	First  string `yaml:"first"`
+	Second string `yaml:"second"`
+}
+
+func TestCollectionInputValidatesWithoutReadingSource(t *testing.T) {
+	input, err := promptkit.CollectionInput("docs", configkit.File(filepath.Join(t.TempDir(), "not-read.yaml")))
+	if err != nil {
+		t.Fatal(err)
+	}
+	if input.Destination.String() != "/collections/docs" || !input.Imports {
+		t.Fatalf("collection input = %+v, want destination /collections/docs with imports enabled", input)
+	}
+	for _, namespace := range []string{"", " ", "docs/api"} {
+		if _, err := promptkit.CollectionInput(namespace, configkit.NoSource()); !errors.Is(err, promptkit.ErrInvalidConfig) {
+			t.Errorf("namespace %q error = %v, want ErrInvalidConfig", namespace, err)
+		}
+	}
+}
+
+func TestCollectionInputLoadsNativeImportsAndLaterOverride(t *testing.T) {
+	directory := t.TempDir()
+	mainPath := filepath.Join(directory, "collection.yaml")
+	defaultsPath := filepath.Join(directory, "defaults.yaml")
+	if err := os.WriteFile(mainPath, []byte("config:\n  imports:\n    - path: defaults.yaml\n      required: true\n"), 0o600); err != nil {
+		t.Fatal(err)
+	}
+	if err := os.WriteFile(defaultsPath, []byte("prompts:\n  - id: greeting\n    messages:\n      - role: user\n        content: Hello\n    output:\n      format: text\n      validation_mode: none\n"), 0o600); err != nil {
+		t.Fatal(err)
+	}
+	input, err := promptkit.CollectionInput("docs", configkit.File(mainPath))
+	if err != nil {
+		t.Fatal(err)
+	}
+	engine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{
+		Underlays: []configkit.Underlay{{Inputs: []configkit.Input{input}}},
+		Main: &configkit.Input{
+			Source:      configkit.Mapping("later collection override", map[string]any{"prompts": []any{map[string]any{"id": "greeting", "inputs": []any{map[string]any{"name": "overridden"}}, "messages": []any{map[string]any{"role": "user", "content": "Updated"}}, "output": map[string]any{"format": "text", "validation_mode": "none"}}}}),
+			Destination: configkit.Keys("collections", "docs"),
+		},
+	}})
+	if err != nil {
+		t.Fatalf("construct engine with native collection import: %v", err)
+	}
+	inspection, err := engine.InspectPrompt(context.Background(), "docs/greeting")
+	if err != nil || inspection.PromptID != "docs/greeting" || len(inspection.Inputs) != 1 || inspection.Inputs[0].Name != "overridden" {
+		t.Fatalf("inspect imported and overlaid prompt = (%+v, %v)", inspection, err)
+	}
+}
+
+func TestCollectionInputLoadsEmbeddedImportsAndLaterOverride(t *testing.T) {
+	files := fstest.MapFS{
+		"collection.yaml": {Data: []byte("config:\n  imports:\n    - path: defaults.yaml\n      required: true\n")},
+		"defaults.yaml":   {Data: []byte("prompts:\n  - id: greeting\n    messages:\n      - role: user\n        content: Hello\n    output:\n      format: text\n      validation_mode: none\n")},
+	}
+	filesystem := configkit.NewFilesystem("embedded collection", files)
+	input, err := promptkit.CollectionInput("docs", filesystem.File("collection.yaml"))
+	if err != nil {
+		t.Fatal(err)
+	}
+	input.Filesystem = &configkit.FilesystemContext{FS: &filesystem, Root: "."}
+	engine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{
+		Underlays: []configkit.Underlay{{Inputs: []configkit.Input{input}}},
+		Main: &configkit.Input{
+			Source:      configkit.Mapping("later collection override", map[string]any{"prompts": []any{map[string]any{"id": "greeting", "inputs": []any{map[string]any{"name": "overridden"}}, "messages": []any{map[string]any{"role": "user", "content": "Updated"}}, "output": map[string]any{"format": "text", "validation_mode": "none"}}}}),
+			Destination: configkit.Keys("collections", "docs"),
+		},
+	}})
+	if err != nil {
+		t.Fatalf("construct engine with embedded collection import: %v", err)
+	}
+	inspection, err := engine.InspectPrompt(context.Background(), "docs/greeting")
+	if err != nil || inspection.PromptID != "docs/greeting" || len(inspection.Inputs) != 1 || inspection.Inputs[0].Name != "overridden" {
+		t.Fatalf("inspect imported and overlaid prompt = (%+v, %v)", inspection, err)
+	}
+}
+
+func TestDirectoryInputsSelectsSortedFilesWithRootContext(t *testing.T) {
+	root := t.TempDir()
+	if err := os.Mkdir(filepath.Join(root, "nested"), 0o755); err != nil {
+		t.Fatal(err)
+	}
+	for name, data := range map[string]string{
+		"a.yaml": "first: one\n", "nested/b.yml": "second: two\n", "nested/body.md": "message",
+	} {
+		if err := os.WriteFile(filepath.Join(root, name), []byte(data), 0o600); err != nil {
+			t.Fatal(err)
+		}
+	}
+	inputs, err := promptkit.DirectoryInputs(root)
+	if err != nil || len(inputs) != 2 || inputs[0].Imports || inputs[1].Imports {
+		t.Fatalf("directory inputs: %+v, %v", inputs, err)
+	}
+	assertSourceInputs(t, inputs, "message")
+	if _, err := promptkit.DirectoryInputs(filepath.Join(root, "missing")); err == nil {
+		t.Fatal("missing selected directory succeeded")
+	}
+	if empty, err := promptkit.DirectoryInputs(t.TempDir()); err != nil || len(empty) != 0 {
+		t.Fatalf("empty directory: %+v, %v", empty, err)
+	}
+}
+
+func TestFilesystemInputsSelectsEmbeddedFilesWithRootContext(t *testing.T) {
+	files := fstest.MapFS{
+		"settings/a.yaml":         {Data: []byte("first: one\n")},
+		"settings/nested/b.yml":   {Data: []byte("second: two\n")},
+		"settings/nested/body.md": {Data: []byte("message")},
+	}
+	inputs, err := promptkit.FilesystemInputs("embedded", files, "settings")
+	if err != nil || len(inputs) != 2 || inputs[0].Imports || inputs[1].Imports {
+		t.Fatalf("filesystem inputs: %+v, %v", inputs, err)
+	}
+	assertSourceInputs(t, inputs, "message")
+	if _, err := promptkit.FilesystemInputs("embedded", files, "missing"); err == nil {
+		t.Fatal("missing selected root succeeded")
+	}
+}
+
+func assertSourceInputs(t *testing.T, inputs []configkit.Input, expected string) {
+	t.Helper()
+	result, err := configkit.LoadLayered(configkit.Input{Source: configkit.NoSource()}, configkit.LayeredOptions[sourceDocument]{
+		Underlays: []configkit.Underlay{{Inputs: inputs}},
+	})
+	if err != nil {
+		t.Fatal(err)
+	}
+	if result.Config.First != "one" || result.Config.Second != "two" {
+		t.Fatalf("configuration files did not compose: %+v", result.Config)
+	}
+	field, found := result.Metadata().Lookup(configkit.Keys("second"))
+	if !found || len(field.Origin.Locations) != 1 {
+		t.Fatalf("missing second-field origin: %+v", field)
+	}
+	context, found := result.SourceContext(field.Origin.Locations[0].SourceID)
+	if !found || context.BaseDir != "nested" {
+		t.Fatalf("unexpected rooted source context: %+v", context)
+	}
+	data, err := fs.ReadFile(context.FS, "nested/body.md")
+	if err != nil || string(data) != expected {
+		t.Fatalf("rooted resource read: %q, %v", data, err)
+	}
+}

construction_contract_test.go

@@ -0,0 +1,317 @@
+package promptkit_test
+
+import (
+	"bytes"
+	"context"
+	"errors"
+	"io"
+	"net/http"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+	"testing/fstest"
+
+	"go.maximumdirect.net/configkit"
+	"go.maximumdirect.net/promptkit"
+)
+
+func TestProfileEnvironmentLookupAcceptsNonShellNames(t *testing.T) {
+	const name = "promptkit.test-api-key"
+	for _, test := range []struct {
+		name, value, direct, want string
+		unset, clear              bool
+	}{
+		{name: "set", value: " synthetic-key ", want: "Bearer synthetic-key"},
+		{name: "empty"},
+		{name: "whitespace", value: "  "},
+		{name: "unset", unset: true},
+		{name: "direct precedence", value: "synthetic-key", direct: "direct-key", want: "Bearer direct-key"},
+		{name: "explicit clear", value: "synthetic-key", clear: true},
+	} {
+		t.Run(test.name, func(t *testing.T) {
+			t.Setenv(name, test.value)
+			if test.unset {
+				if err := os.Unsetenv(name); err != nil {
+					t.Fatal(err)
+				}
+			}
+			cfg := simpleConfig("hello")
+			cfg.Sources.Underlays = []configkit.Underlay{{Inputs: []configkit.Input{*cfg.Sources.Main}}}
+			endpoint, model := "http://example.test/v1", "model"
+			lowerName := name
+			cfg.Sources.Underlays = append(cfg.Sources.Underlays, configkit.Underlay{Inputs: []configkit.Input{{
+				Source: promptkit.ProfileSource("default credentials", []promptkit.ProfileConfig{
+					{ID: "profile", Endpoint: &endpoint, Model: &model, APIKeyEnv: &lowerName},
+				}),
+			}}})
+			env := " " + name + " "
+			if test.clear {
+				env = ""
+			}
+			cfg.Sources.Main = &configkit.Input{Source: promptkit.ProfileSource("credentials", []promptkit.ProfileConfig{
+				{ID: "profile", Endpoint: &endpoint, Model: &model, APIKeyEnv: &env},
+			})}
+			var calls int
+			var auth string
+			cfg.HTTPClient = &http.Client{Transport: roundTripFunc(func(request *http.Request) (*http.Response, error) {
+				calls++
+				auth = request.Header.Get("Authorization")
+				return &http.Response{StatusCode: http.StatusOK, Header: make(http.Header), Body: io.NopCloser(strings.NewReader(`{"choices":[{"message":{"content":"ok"}}]}`))}, nil
+			})}
+			engine, err := promptkit.NewEngine(cfg)
+			if err != nil {
+				t.Fatal(err)
+			}
+			_, err = engine.Run(context.Background(), promptkit.RunRequest{PromptID: "prompt", APIKey: test.direct})
+			if err != nil || calls != 1 || auth != test.want {
+				t.Fatalf("run: err=%v calls=%d authorization=%q, want %q", err, calls, auth, test.want)
+			}
+		})
+	}
+}
+
+func TestPublicConstructionWithLocalBackendAndProgrammaticProfile(t *testing.T) {
+	backendID, model := "workstation", "offline-model"
+	prompt := configkit.Input{Source: configkit.Mapping("prompt", map[string]any{
+		"prompts": []any{map[string]any{
+			"id":              "hello",
+			"default_profile": "offline",
+			"messages":        []any{map[string]any{"role": "user", "content": "Hello"}},
+			"output":          map[string]any{"format": "text", "validation_mode": "none"},
+		}},
+	})}
+	client := &fakeLLMClient{response: &promptkit.GenerateResponse{Content: "reply"}}
+	localBackend := promptkit.LocalBackend("http://localhost:8000/v1", 2)
+	localBackend.ID = backendID
+	engine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{
+		Underlays: []configkit.Underlay{{Inputs: []configkit.Input{
+			{Source: promptkit.BackendSource("local backend", []promptkit.BackendConfig{
+				localBackend,
+			})},
+			{Source: promptkit.ProfileSource("profile", []promptkit.ProfileConfig{
+				{ID: "offline", Backend: &backendID, Model: &model},
+			})},
+		}}},
+		Main: &prompt,
+	}}, promptkit.WithLLMClient(client))
+	if err != nil {
+		t.Fatalf("construct engine: %v", err)
+	}
+	result, err := engine.Run(context.Background(), promptkit.RunRequest{PromptID: "hello"})
+	if err != nil {
+		t.Fatalf("run prompt: %v", err)
+	}
+	if result.RawOutput != "reply" || result.SelectedBackendID != backendID || result.EffectiveModelParams.Model != model {
+		t.Fatalf("unexpected result: %+v", result)
+	}
+	if len(client.requests) != 1 || client.requests[0].Target.Endpoint != "http://localhost:8000/v1" {
+		t.Fatalf("unexpected model request: %+v", client.requests)
+	}
+}
+
+func TestFormerCatalogBackendIDsWorkWhenConfigured(t *testing.T) {
+	openRouterID, rakestrawhomeID := promptkit.BackendOpenRouter, promptkit.BackendRakestrawHome
+	openRouterEndpoint, rakestrawhomeEndpoint := "https://openrouter.example/v1", "https://inference.example/v1"
+	model := "configured-model"
+	engine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{
+		Underlays: []configkit.Underlay{{Inputs: []configkit.Input{
+			{Source: promptkit.BackendSource("application backends", []promptkit.BackendConfig{
+				{ID: openRouterID, Endpoint: &openRouterEndpoint},
+				{ID: rakestrawhomeID, Endpoint: &rakestrawhomeEndpoint},
+			})},
+			{Source: promptkit.ProfileSource("application profiles", []promptkit.ProfileConfig{
+				{ID: "openrouter-profile", Backend: &openRouterID, Model: &model},
+				{ID: "rakestrawhome-profile", Backend: &rakestrawhomeID, Model: &model},
+			})},
+		}}},
+	}})
+	if err != nil {
+		t.Fatalf("construct engine with explicitly configured provider IDs: %v", err)
+	}
+	for profileID, backendID := range map[string]string{
+		"openrouter-profile":    openRouterID,
+		"rakestrawhome-profile": rakestrawhomeID,
+	} {
+		inspection, err := engine.InspectProfile(context.Background(), profileID)
+		if err != nil || inspection.EffectiveModelParams.BackendID != backendID {
+			t.Fatalf("inspect %q = (%+v, %v), want backend %q", profileID, inspection, err, backendID)
+		}
+	}
+}
+
+func TestPublicConstructionCombinesEmbeddedOperatorAndProgrammaticInputs(t *testing.T) {
+	embedded, err := promptkit.FilesystemInputs("embedded defaults", fstest.MapFS{
+		"config/default.yaml": &fstest.MapFile{Data: []byte(`
+profiles:
+  - id: application
+    endpoint: https://example.invalid/v1
+    model: default-model
+prompts:
+  - id: hello
+    default_profile: application
+    messages:
+      - role: user
+        content: Hello
+    output:
+      format: text
+      validation_mode: none
+`)},
+	}, "config")
+	if err != nil {
+		t.Fatalf("select embedded defaults: %v", err)
+	}
+	operatorDirectory := t.TempDir()
+	if err := os.WriteFile(filepath.Join(operatorDirectory, "override.yaml"), []byte("profiles:\n  - id: application\n    endpoint: https://example.invalid/v1\n    model: operator-model\n"), 0o644); err != nil {
+		t.Fatalf("write operator input: %v", err)
+	}
+	operator, err := promptkit.DirectoryInputs(operatorDirectory)
+	if err != nil {
+		t.Fatalf("select operator inputs: %v", err)
+	}
+	programmaticModel, programmaticEndpoint := "programmatic-model", "https://example.invalid/v1"
+	programmatic := configkit.Input{Source: promptkit.ProfileSource("programmatic settings", []promptkit.ProfileConfig{
+		{ID: "application", Endpoint: &programmaticEndpoint, Model: &programmaticModel},
+	})}
+	for _, test := range []struct {
+		name  string
+		apply bool
+		want  string
+	}{
+		{name: "embedded", want: "default-model"},
+		{name: "operator", apply: true, want: "operator-model"},
+		{name: "programmatic", apply: true, want: "programmatic-model"},
+	} {
+		t.Run(test.name, func(t *testing.T) {
+			underlays := []configkit.Underlay{{Inputs: embedded}}
+			if test.apply {
+				underlays = append(underlays, configkit.Underlay{Inputs: operator})
+			}
+			if test.name == "programmatic" {
+				underlays = append(underlays, configkit.Underlay{Inputs: []configkit.Input{programmatic}})
+			}
+			engine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{Underlays: underlays}})
+			if err != nil {
+				t.Fatalf("construct engine: %v", err)
+			}
+			prepared, err := engine.Prepare(context.Background(), promptkit.RunRequest{PromptID: "hello"})
+			if err != nil {
+				t.Fatalf("prepare prompt: %v", err)
+			}
+			if prepared.EffectiveModelParams.Model != test.want {
+				t.Fatalf("model = %q, want %q", prepared.EffectiveModelParams.Model, test.want)
+			}
+		})
+	}
+}
+
+func TestPublicConstructionPreservesConfigKitLoadError(t *testing.T) {
+	engine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{Main: &configkit.Input{
+		Source: configkit.Mapping("invalid", map[string]any{"profiles": []any{map[string]any{"id": "broken", "unknown": true}}}),
+	}}})
+	var loadErr *configkit.LoadError
+	if engine != nil || !errors.Is(err, promptkit.ErrInvalidConfig) || !errors.As(err, &loadErr) {
+		t.Fatalf("construction = (%#v, %v), want invalid config preserving LoadError", engine, err)
+	}
+}
+
+func TestPublicConstructionRejectsSuppliedInvalidMainInput(t *testing.T) {
+	main := configkit.Input{}
+	engine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{Main: &main}})
+	if engine != nil || !errors.Is(err, promptkit.ErrInvalidConfig) {
+		t.Fatalf("construction = (%#v, %v), want invalid supplied main input", engine, err)
+	}
+}
+
+func TestPublicConstructionUsesProgrammaticBackendAndProfileSources(t *testing.T) {
+	endpoint, backendID, model := "http://application.example/v1", "application-backend", "synthetic-model"
+	engine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{
+		Underlays: []configkit.Underlay{{Inputs: []configkit.Input{
+			{Source: promptkit.BackendSource("operator backend", []promptkit.BackendConfig{
+				{ID: backendID, Endpoint: &endpoint},
+			})},
+			{Source: promptkit.ProfileSource("operator profile", []promptkit.ProfileConfig{
+				{ID: "application-profile", Backend: &backendID, Model: &model},
+			})},
+		}}},
+	}})
+	if err != nil {
+		t.Fatalf("construct engine with backend overlay: %v", err)
+	}
+	inspection, err := engine.InspectProfile(context.Background(), "application-profile")
+	if err != nil {
+		t.Fatalf("inspect configured profile: %v", err)
+	}
+	if inspection.EffectiveModelParams.Endpoint != endpoint {
+		t.Fatalf("effective endpoint = %q, want %q", inspection.EffectiveModelParams.Endpoint, endpoint)
+	}
+}
+
+func TestBackendSourceOwnsSmokeTestValuesAndDefersDefaultLookup(t *testing.T) {
+	backendEndpoint := "https://backend.example/v1"
+	defaultProfile := "not-yet-listed"
+	reachabilityEndpoint := " https://probe.example/ready?check=1 "
+	authenticationKind := " OPENROUTER_KEY "
+	authenticationEndpoint := "https://probe.example/key?scope=current"
+	source := promptkit.BackendSource("diagnostic backend", []promptkit.BackendConfig{
+		{
+			ID:       "local",
+			Endpoint: &backendEndpoint,
+			SmokeTests: &promptkit.BackendSmokeTestConfig{
+				DefaultProfile: &defaultProfile,
+				Reachability:   &promptkit.ReachabilityProbeConfig{Endpoint: &reachabilityEndpoint},
+				Authentication: &promptkit.AuthenticationProbeConfig{
+					Kind: &authenticationKind, Endpoint: &authenticationEndpoint,
+				},
+			},
+		},
+	})
+	backendEndpoint = "https://mutated.example/v1"
+	defaultProfile = "mutated-profile"
+	reachabilityEndpoint = "https://mutated.example/ready"
+	authenticationKind = "none"
+	authenticationEndpoint = "https://mutated.example/key"
+
+	engine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{
+		Underlays: []configkit.Underlay{{Inputs: []configkit.Input{{Source: source}}}},
+	}})
+	if err != nil {
+		t.Fatalf("construct engine with dangling diagnostic default: %v", err)
+	}
+	inspection, err := engine.InspectConfiguration()
+	if err != nil {
+		t.Fatal(err)
+	}
+	var effective bytes.Buffer
+	if err := inspection.WriteEffective(&effective); err != nil {
+		t.Fatal(err)
+	}
+	contents := effective.String()
+	for _, want := range []string{
+		"https://backend.example/v1",
+		"not-yet-listed",
+		"https://probe.example/ready?check=1",
+		"openrouter_key",
+		"https://probe.example/key?scope=current",
+	} {
+		if !strings.Contains(contents, want) {
+			t.Errorf("effective configuration does not contain %q:\n%s", want, contents)
+		}
+	}
+	for _, unwanted := range []string{
+		"mutated.example",
+		"mutated-profile",
+	} {
+		if strings.Contains(contents, unwanted) {
+			t.Errorf("effective configuration unexpectedly contains %q:\n%s", unwanted, contents)
+		}
+	}
+}
+
+func TestPreparationSelectsSinglePromptDefinitionByID(t *testing.T) {
+	engine := newSimpleEngine(t, "selected")
+	prepared, err := engine.Prepare(context.Background(), promptkit.RunRequest{PromptID: "prompt"})
+	if err != nil || prepared == nil || len(prepared.Messages) != 1 || prepared.Messages[0].Content != "selected" {
+		t.Fatalf("prepare prompt by ID = (%+v, %v)", prepared, err)
+	}
+}

consumer_test_helpers_test.go

@@ -0,0 +1,73 @@
+package promptkit_test
+
+import (
+	"context"
+	"fmt"
+	"net/http"
+	"sync/atomic"
+	"testing"
+
+	"go.maximumdirect.net/configkit"
+	"go.maximumdirect.net/promptkit"
+)
+
+type fakeLLMClient struct {
+	response  *promptkit.GenerateResponse
+	responses []*promptkit.GenerateResponse
+	err       error
+	requests  []promptkit.GenerateRequest
+}
+
+func (f *fakeLLMClient) Generate(_ context.Context, request promptkit.GenerateRequest) (*promptkit.GenerateResponse, error) {
+	f.requests = append(f.requests, request)
+	if f.err != nil {
+		return nil, f.err
+	}
+	if len(f.responses) > 0 {
+		index := len(f.requests) - 1
+		if index >= len(f.responses) {
+			return nil, fmt.Errorf("no response configured for generation %d", index+1)
+		}
+		return f.responses[index], nil
+	}
+	return f.response, nil
+}
+
+type countingLLMClient struct{ calls *atomic.Int64 }
+
+func (client countingLLMClient) Generate(context.Context, promptkit.GenerateRequest) (*promptkit.GenerateResponse, error) {
+	if client.calls != nil {
+		client.calls.Add(1)
+	}
+	return &promptkit.GenerateResponse{Content: "ok"}, nil
+}
+
+type roundTripFunc func(*http.Request) (*http.Response, error)
+
+func (f roundTripFunc) RoundTrip(request *http.Request) (*http.Response, error) { return f(request) }
+
+// newSimpleEngine exercises the public constructor with a complete keyed
+// document. Tests that need layering or resources select their own inputs.
+func newSimpleEngine(t *testing.T, content string, options ...promptkit.Option) *promptkit.Engine {
+	t.Helper()
+	engine, err := promptkit.NewEngine(simpleConfig(content), options...)
+	if err != nil {
+		t.Fatalf("construct consumer test engine: %v", err)
+	}
+	return engine
+}
+
+func simpleConfig(content string) promptkit.Config {
+	return promptkit.Config{Sources: promptkit.ConfigurationSources{
+		Main: &configkit.Input{Source: configkit.Mapping("consumer test", map[string]any{
+			"profiles": []any{map[string]any{"id": "profile",
+				"endpoint": "http://example.test/v1", "model": "model",
+			}},
+			"prompts": []any{map[string]any{"id": "prompt",
+				"default_profile": "profile",
+				"messages":        []any{map[string]any{"role": "user", "content": content}},
+				"output":          map[string]any{"format": "text", "validation_mode": "none"},
+			}},
+		})},
+	}}
+}

convert.go

@@ -1,10 +1,12 @@
 package promptkit
 
 import (
+	"fmt"
 	"reflect"
+	"unicode/utf8"
 
-	"gitea.maximumdirect.net/eric/promptkit/internal/domain"
-	"gitea.maximumdirect.net/eric/promptkit/internal/jsonvalue"
+	"go.maximumdirect.net/promptkit/internal/domain"
+	"go.maximumdirect.net/promptkit/internal/jsonvalue"
 )
 
 func toDomainRunRequest(req RunRequest) (domain.RunRequest, error) {
@@ -12,26 +14,64 @@ func toDomainRunRequest(req RunRequest) (domain.RunRequest, error) {
 	if err != nil {
 		return domain.RunRequest{}, err
 	}
+	appendedMessages, err := toDomainAppendedMessages(req.AppendedMessages)
+	if err != nil {
+		return domain.RunRequest{}, err
+	}
+	var observer domain.ExecutionObserver
+	if req.Observer != nil {
+		callback := req.Observer
+		observer = func(event domain.ExecutionEvent) { callback(fromDomainExecutionEvent(event)) }
+	}
 	return domain.RunRequest{
-		PromptID:      req.PromptID,
-		PromptVersion: req.PromptVersion,
-		ProfileID:     req.ProfileID,
-		SessionID:     req.SessionID,
-		APIKey:        req.APIKey,
-		Inputs:        toDomainArtifactRefMap(req.Inputs),
-		Vars:          copyStringMap(req.Vars),
-		Execution:     execution,
-		Validation:    toDomainOutputContractPtr(req.Validation),
+		PromptID:         req.PromptID,
+		ProfileID:        req.ProfileID,
+		SessionID:        req.SessionID,
+		APIKey:           req.APIKey,
+		Inputs:           toDomainArtifactRefMap(req.Inputs),
+		Vars:             copyStringMap(req.Vars),
+		Execution:        execution,
+		Validation:       toDomainOutputContractPtr(req.Validation),
+		AppendedMessages: appendedMessages,
+		TransientRetries: copyIntPtr(req.TransientRetries),
+		Observer:         observer,
 	}, nil
 }
 
+func toDomainAppendedMessages(messages []RenderedMessage) ([]domain.RenderedMessage, error) {
+	if messages == nil {
+		return nil, nil
+	}
+
+	converted := make([]domain.RenderedMessage, len(messages))
+	for index, message := range messages {
+		if !utf8.ValidString(message.Content) {
+			return nil, fmt.Errorf("appended message %d content must be valid UTF-8", index)
+		}
+		role, err := domain.NormalizeMessageRole(message.Role)
+		if err != nil {
+			return nil, fmt.Errorf("appended message %d role: %w", index, err)
+		}
+
+		cacheControl, err := domain.NormalizeCacheControl(toDomainCacheControl(message.CacheControl))
+		if err != nil {
+			return nil, fmt.Errorf("appended message %d cache_control: %w", index, err)
+		}
+		converted[index] = domain.RenderedMessage{
+			Role:         role,
+			Content:      message.Content,
+			CacheControl: cacheControl,
+		}
+	}
+	return converted, nil
+}
+
 func fromDomainPreparedRun(prepared *domain.PreparedRun) *PreparedRun {
 	if prepared == nil {
 		return nil
 	}
 	return &PreparedRun{
 		PromptID:             prepared.PromptID,
-		PromptVersion:        prepared.PromptVersion,
 		PromptHash:           prepared.PromptHash,
 		SelectedProfileID:    prepared.SelectedProfileID,
 		SelectedBackendID:    prepared.SelectedBackendID,
@@ -58,7 +98,6 @@ func fromDomainRunResult(result *domain.RunResult) *RunResult {
 		RawOutput:            result.RawOutput,
 		Validation:           fromDomainValidationResult(result.Validation),
 		PromptID:             result.PromptID,
-		PromptVersion:        result.PromptVersion,
 		PromptHash:           result.PromptHash,
 		SessionID:            result.SessionID,
 		RenderedPromptHash:   result.RenderedPromptHash,
@@ -90,8 +129,22 @@ func toDomainGenerateResponse(resp *GenerateResponse) *domain.GenerateResponse {
 		return nil
 	}
 	return &domain.GenerateResponse{
-		Content: resp.Content,
-		Usage:   toDomainTokenUsage(resp.Usage),
+		Content:       resp.Content,
+		Usage:         toDomainTokenUsage(resp.Usage),
+		UsagePresence: toDomainTokenUsagePresence(resp.UsagePresence),
+	}
+}
+
+func toDomainTokenUsagePresence(presence *TokenUsagePresence) *domain.TokenUsagePresence {
+	if presence == nil {
+		return nil
+	}
+	return &domain.TokenUsagePresence{
+		PromptTokens:     presence.PromptTokens,
+		CompletionTokens: presence.CompletionTokens,
+		TotalTokens:      presence.TotalTokens,
+		CachedTokens:     presence.CachedTokens,
+		CacheWriteTokens: presence.CacheWriteTokens,
 	}
 }
 
@@ -155,6 +208,12 @@ func toDomainExecutionTargetOverride(override *ExecutionTargetOverride) (*domain
 }
 
 func fromDomainExecutionTarget(target domain.ExecutionTarget) ExecutionTarget {
+	return fromDomainExecutionTargetWithOwnedParams(target, copyAnyMap(target.ExtraParams))
+}
+
+// fromDomainExecutionTargetWithOwnedParams projects scalar settings and takes
+// ownership of extraParams, which the caller has already copied.
+func fromDomainExecutionTargetWithOwnedParams(target domain.ExecutionTarget, extraParams map[string]any) ExecutionTarget {
 	return ExecutionTarget{
 		BackendID:       target.BackendID,
 		Endpoint:        target.Endpoint,
@@ -166,7 +225,7 @@ func fromDomainExecutionTarget(target domain.ExecutionTarget) ExecutionTarget {
 		ServiceTier:     target.ServiceTier,
 		ReasoningEffort: target.ReasoningEffort,
 		APIKeyEnv:       target.APIKeyEnv,
-		ExtraParams:     copyAnyMap(target.ExtraParams),
+		ExtraParams:     extraParams,
 	}
 }
 
@@ -196,7 +255,6 @@ func fromDomainPromptInspection(inspection *domain.PromptInspection) *PromptInsp
 	}
 	return &PromptInspection{
 		PromptID:         inspection.PromptID,
-		PromptVersion:    inspection.PromptVersion,
 		PromptHash:       inspection.PromptHash,
 		DefaultProfileID: inspection.DefaultProfileID,
 		Inputs:           inputs,
@@ -285,6 +343,16 @@ func fromDomainRenderedMessages(messages []domain.RenderedMessage) []RenderedMes
 	return out
 }
 
+func toDomainCacheControl(cacheControl *CacheControl) *domain.CacheControl {
+	if cacheControl == nil {
+		return nil
+	}
+	return &domain.CacheControl{
+		Type: domain.CacheControlType(cacheControl.Type),
+		TTL:  cacheControl.TTL,
+	}
+}
+
 func fromDomainCacheControl(cacheControl *domain.CacheControl) *CacheControl {
 	if cacheControl == nil {
 		return nil

convert_appended_messages_test.go

@@ -0,0 +1,108 @@
+package promptkit
+
+import (
+	"reflect"
+	"strings"
+	"testing"
+
+	"go.maximumdirect.net/promptkit/internal/domain"
+)
+
+func TestToDomainAppendedMessages(t *testing.T) {
+	tests := []struct {
+		name        string
+		messages    []RenderedMessage
+		want        []domain.RenderedMessage
+		wantErr     string
+		privateRole string
+		privateText string
+	}{
+		{
+			name: "normalizes supported roles",
+			messages: []RenderedMessage{
+				{Role: " DeVeLoPeR ", Content: "developer"},
+				{Role: "SyStEm", Content: "system"},
+				{Role: "\tUsEr\n", Content: "user"},
+				{Role: "assistant", Content: "assistant"},
+			},
+			want: []domain.RenderedMessage{
+				{Role: domain.RoleDeveloper, Content: "developer"},
+				{Role: domain.RoleSystem, Content: "system"},
+				{Role: domain.RoleUser, Content: "user"},
+				{Role: domain.RoleAssistant, Content: "assistant"},
+			},
+		},
+		{
+			name:     "invalid role UTF-8",
+			messages: []RenderedMessage{{Role: string([]byte{0xff}), Content: "private-content"}},
+			wantErr:  "appended message 0 role",
+		},
+		{
+			name:     "invalid content UTF-8",
+			messages: []RenderedMessage{{Role: "private-role", Content: string([]byte{0xff})}},
+			wantErr:  "appended message 0 content",
+		},
+		{
+			name:        "unsupported role",
+			messages:    []RenderedMessage{{Role: "private-role", Content: "private-content"}},
+			wantErr:     "appended message 0 role",
+			privateRole: "private-role",
+			privateText: "private-content",
+		},
+		{
+			name:        "invalid cache control",
+			messages:    []RenderedMessage{{Role: RoleUser, Content: "private-content", CacheControl: &CacheControl{Type: "private-cache"}}},
+			wantErr:     "appended message 0 cache_control",
+			privateText: "private-content",
+		},
+		{
+			name: "empty and whitespace content",
+			messages: []RenderedMessage{
+				{Role: RoleUser, Content: ""},
+				{Role: RoleAssistant, Content: " \t\n "},
+			},
+			want: []domain.RenderedMessage{
+				{Role: domain.RoleUser, Content: ""},
+				{Role: domain.RoleAssistant, Content: " \t\n "},
+			},
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			got, err := toDomainAppendedMessages(test.messages)
+			if test.wantErr != "" {
+				if err == nil || !strings.Contains(err.Error(), test.wantErr) {
+					t.Fatalf("error = %v, want %q", err, test.wantErr)
+				}
+				for _, privateValue := range []string{test.privateRole, test.privateText} {
+					if privateValue != "" && strings.Contains(err.Error(), privateValue) {
+						t.Fatalf("error exposed appended message data %q: %v", privateValue, err)
+					}
+				}
+				return
+			}
+			if err != nil {
+				t.Fatalf("toDomainAppendedMessages() error = %v", err)
+			}
+			if !reflect.DeepEqual(got, test.want) {
+				t.Fatalf("toDomainAppendedMessages() = %#v, want %#v", got, test.want)
+			}
+		})
+	}
+}
+
+func TestToDomainAppendedMessagesCopiesCacheControl(t *testing.T) {
+	cacheControl := &CacheControl{Type: CacheControlEphemeral, TTL: "1h"}
+	messages := []RenderedMessage{{Role: RoleUser, Content: "content", CacheControl: cacheControl}}
+	request, err := toDomainRunRequest(RunRequest{AppendedMessages: messages})
+	if err != nil {
+		t.Fatalf("toDomainRunRequest() error = %v", err)
+	}
+
+	messages[0].Content = "changed"
+	cacheControl.TTL = ""
+	if request.AppendedMessages[0].Content != "content" || request.AppendedMessages[0].CacheControl.TTL != "1h" {
+		t.Fatalf("domain request did not retain an independent appended-message copy: %#v", request.AppendedMessages)
+	}
+}

doc.go

@@ -1,27 +1,32 @@
 // Package promptkit provides an embeddable engine for preparing and executing
 // prompt-defined LLM workflows.
 //
-// Applications construct an [Engine] with [NewEngine], select filesystem or
-// in-memory sources and optional engine-scoped [Backend] registrations, and
-// call [Engine.InspectPrompt], [Engine.InspectProfile], [Engine.Prepare],
-// [Engine.PrepareExecution], [Engine.Run], or [Engine.RunPrepared]. Concrete
-// registries, repositories, validators, and the built-in OpenAI-compatible
-// client remain internal implementation details.
+// Applications construct an [Engine] with [NewEngine], select ConfigKit
+// sources for prompts, profiles, and backends, and
+// call [Engine.InspectConfiguration], [Engine.InspectPrompt],
+// [Engine.InspectRequirements], [Engine.InspectProfile], [Engine.ListProfiles],
+// [Engine.ListPrompts], [Engine.Prepare], [Engine.PrepareExecution],
+// [Engine.Run], or [Engine.RunPrepared]. Concrete registries, repositories,
+// validators, and the built-in OpenAI-compatible client remain internal
+// implementation details.
 //
 // # Concurrency and ownership
 //
-// An Engine supports concurrent InspectPrompt, InspectProfile, Prepare,
+// An Engine supports concurrent InspectConfiguration, InspectPrompt,
+// InspectRequirements, InspectProfile, ListProfiles, ListPrompts, Prepare,
 // PrepareExecution, Run, and RunPrepared calls. Engine-local backend policies
 // bound admitted Run and RunPrepared calls and model generations where
 // configured, while different backend pools and unlimited backends continue
 // independently. An injected [LLMClient] or [ArtifactReader] can therefore
 // still receive concurrent calls and must be safe for that use.
 //
-// NewEngine copies in-memory profiles and backend definitions. Prepare,
-// PrepareExecution, and Run copy request maps, slices, pointer values, and
-// JSON-compatible extra parameters before using them. InspectPrompt and
-// InspectProfile return copied inspection values. Returned values and values
-// passed to extension interfaces are likewise isolated from engine state.
+// ConfigKit captures the selected configuration when NewEngine constructs an
+// Engine. Prepare, PrepareExecution, and Run copy request maps, slices, pointer
+// values, and JSON-compatible extra parameters before using them. InspectPrompt,
+// InspectRequirements, and InspectProfile return copied inspection values.
+// ListProfiles and ListPrompts return caller-owned identity slices.
+// Returned values and values passed to extension interfaces are likewise
+// isolated from engine state.
 // Callers own those copies and may mutate them after the call that supplied or
 // returned them. [CapacityError] values are caller-owned and may be mutated
 // without affecting engine state or another error. Immutable [GenerationError]
@@ -52,10 +57,13 @@
 // used by those values.
 //
 // Construction, inspection, handle, and error values, including [Config],
-// [Backend], [RunRequest], [ArtifactRef], [ExecutionTargetOverride], [Profile],
-// [OpenAICompatibleProfileConfig], [ProfileInspection],
-// [PromptInputDefinition], [PromptInspection], [PreparedExecution],
-// [CapacityError], and [GenerationError], do not have stable JSON
+// [ConfigurationSources], [ConfigurationInspection], [BackendConfig],
+// [ProfileConfig], [RunRequest],
+// [ArtifactRef], [ExecutionTargetOverride], [ProfileInspection],
+// [PromptIdentity], [PromptInputDefinition], [PromptInspection],
+// [RequirementInspectionRequest], [RequirementInspection], [PreparedExecution],
+// [CapacityError], [GenerationError], [RetryExhaustedError], and
+// [ExecutionEvent], do not have stable JSON
 // representations. Direct API keys are nevertheless excluded from JSON for
 // every public value.
 

docs/adr/0001-configuration-boundary.md

@@ -0,0 +1,60 @@
+# 0001: Configuration Ownership At The ConfigKit Handoff
+
+## Status
+
+Accepted.
+
+## Date
+
+2026-09-30
+
+## Context
+
+PromptKit is adopting ConfigKit to standardize configuration semantics and
+simplify its loading machinery. Configuration includes prompts, profiles, and
+backend definitions contributed by maintained catalogs and consuming applications.
+ConfigKit supplies loading, composition, precedence, inheritance, and inspection
+capabilities. PromptKit supplies LLM-specific interpretation and execution.
+
+A clear handoff assigns each library responsibility for the configuration it
+owns and keeps validation aligned with the values an application actually uses.
+
+## Decision
+
+ConfigKit owns configuration processing up to the handoff of the final effective
+configuration. PromptKit takes ownership at that handoff and applies its domain
+interpretation and validation to the final configuration.
+
+The handoff includes effective presence/null information and resource context
+supplied by ConfigKit. PromptKit uses this information to interpret effective
+settings and resolve referenced resources.
+
+PromptKit passes through ConfigKit's inspection capabilities for explicit
+operator use. Configuration history and source explanations serve that
+inspection purpose; domain validation operates on the final effective values.
+
+This boundary applies uniformly to application configuration and maintained
+catalog configuration, including backend definitions.
+
+## Alternatives Considered
+
+- **Independent source or catalog preflight validation:** preserves checks on
+  individual contributions, but duplicates configuration processing and evaluates
+  values that may be replaced before handoff. Raw-source and intermediate-state
+  validation are outside PromptKit's scope, including equivalent test-only checks.
+- **History-based source auditing:** can identify overwritten contributions, but
+  couples domain enforcement to configuration history. Historical inspection is
+  an operator capability rather than a validation mechanism.
+
+## Consequences
+
+PromptKit's configuration logic and tests focus on the final effective values,
+their domain meaning, and resulting execution behavior. ConfigKit owns failures
+arising during its processing, which PromptKit preserves through its error
+boundary. Catalog maintainers own their repositories' publishing checks.
+
+The adoption work routes backend configuration through the same handoff and
+retires PromptKit's source-level and intermediate-configuration validation.
+Effective resource interpretation and explicit operator inspection remain
+supported. This concentrates configuration semantics in ConfigKit and reduces
+duplicated machinery and maintenance in PromptKit.

docs/consumers/configuration-migration.md

@@ -0,0 +1,231 @@
+# Configuration And API Migration
+
+This breaking pre-v1 minor migration adopts Promptkit's vanity module path,
+changes the v0.16-era keyed resources to identified lists, adopts ConfigKit v1,
+and removes prompt-version labels.
+Promptkit installs a default-on bundled OpenRouter catalog; applications choose configuration
+inputs explicitly, and execution selects one prompt by its exact ID. This guide
+explains the shape changes and decisions an application must make; it is not an
+automatic migration tool. The [framework format reference](../formats.md) owns
+the current configuration contract, and the root package GoDoc owns exact API
+signatures and errors.
+
+## Update Go Imports
+
+Promptkit's module and public package import path is now
+`go.maximumdirect.net/promptkit`, replacing
+`gitea.maximumdirect.net/eric/promptkit`. Update application, library, test, and
+generated imports, then select a release using the new declaration:
+
+```sh
+go get go.maximumdirect.net/promptkit@VERSION
+go mod tidy
+```
+
+Replace `VERSION` with a release that declares the vanity module path.
+Earlier tags, including v0.16.0, retain their original module declaration and
+use the former import path. No permanent `replace` directive is needed for the
+cutover. Source browser, licensing, and hosted-release links continue to use
+the Forge repository URLs, independently of Go's module identifier.
+
+The package name remains `promptkit`. The path change itself does not rename
+exported APIs or change their behavior, but Go treats packages imported through
+the two paths as distinct. Update dependent packages that exchange Promptkit
+types together and review any old-path dependencies retained after tidying.
+The configuration and API adaptations below are also part of this migration.
+
+## Select And Compose The Configuration You Need
+
+Supply each backend, profile, prompt, schema, and collection source that the
+application intends to use. Promptkit does not discover directories or install
+provider definitions implicitly. An application may select an explicit
+collection input with `CollectionInput`:
+
+```go
+input, err := promptkit.CollectionInput(
+	"openrouter",
+	configkit.File("config/openrouter.yaml"),
+)
+if err != nil {
+	return err
+}
+engine, err := promptkit.NewEngine(promptkit.Config{
+	Sources: promptkit.ConfigurationSources{
+		Underlays: []configkit.Underlay{{Inputs: []configkit.Input{input}}},
+	},
+})
+```
+
+Collection resource IDs are local IDs, while their runtime identities are
+qualified. For example, a profile with `id: gemini-flash-latest` in collection
+`openrouter` is selected as `openrouter/gemini-flash-latest`. Stored backend,
+prompt-default, and other resource references must use their exact intended IDs;
+Promptkit does not rewrite references when a source is mounted. `extends`
+selects a sibling ID in the same list, not a cross-collection reference.
+Review collisions and all references when moving resources into collections.
+
+The optional source-root `version` header now accepts `2` or `"2"`; headerless
+sources remain valid. A collection's `release_version` remains an optional
+label. Prompt-version labels have been removed: choose one definition per ID
+and store any application-specific release label outside Promptkit. In a
+standalone imported prompt item, root `version` is a source header, not a label.
+
+For example, change a version-keyed definition:
+
+```yaml
+prompts:
+  report:
+    "2.0":
+      messages:
+        - role: user
+          content: Review this report.
+      output:
+        format: text
+        validation_mode: none
+```
+
+to one identified definition:
+
+```yaml
+prompts:
+  - id: report
+    messages:
+      - role: user
+        content: Review this report.
+    output:
+      format: text
+      validation_mode: none
+```
+
+The same conversion applies to backends and profiles. Sibling inheritance uses
+identity references, and a later same-ID record replaces the whole lower record:
+
+```yaml
+profiles:
+  - id: base
+    endpoint: https://example.invalid/v1
+    model: example-model
+  - id: fast
+    extends: {id: base}
+    timeout_seconds: 30
+```
+
+To customize a lower resource's fields, add a distinct inherited variant as
+above; when replacing the same ID, supply the complete intended record. Generic
+append, replace, imports, and inheritance are described by the
+[ConfigKit v1 configuration contract](https://forge.maximumdirect.net/base-layer/configkit/src/tag/v1.0.0/docs/config.md).
+Source-root controls are now `config:`, not `configkit:`. Imports are optional
+unless the selected import declares `required: true`.
+
+Update ConfigKit imports to `go.maximumdirect.net/configkit` at v1.0.0: its Go
+types are incompatible with the old module path. Promptkit now requires Go
+1.27.0 or newer. Source builders take ordered slices with explicit IDs:
+
+```go
+endpoint, model := "https://example.invalid/v1", "example-model"
+base := "base"
+source := promptkit.ProfileSource("application", []promptkit.ProfileConfig{
+	{ID: "base", Endpoint: &endpoint, Model: &model},
+	{ID: "fast", Extends: &base},
+})
+```
+
+Replace former maps passed to `BackendSource` in the same way, setting each
+`BackendConfig.ID`. Duplicates are retained for loading to reject. Resource
+inspection paths now address list indexes, for example
+`configkit.Keys("profiles").Index(0).Key("model")`. Configuration inspection
+and diffs expose resolved values, recorded normalization and provenance.
+`InspectProfile` additionally applies backend completion; request preparation
+can apply further overrides, so the views are not promised to be identical.
+
+Remove `PromptVersion` and `PromptIdentity.Version` uses and the corresponding
+`prompt_version` JSON members. Prompt hashes are opaque equality values and may
+change when the removed field no longer participates in hashing.
+
+## Identity Mapping
+
+The reviewed catalog identities have slash-free profile keys. Their documented
+mapping is:
+
+| Previous identity | Namespaced identity or action |
+| --- | --- |
+| Backend `openrouter` | `openrouter/default` |
+| Backend `rakestrawhome` | `rakestrawhome/default` |
+| Each OpenRouter catalog profile `p` | `openrouter/p`, retaining the exact local key `p` |
+| Profile `rakestrawhome-gemma-4-31b` | `rakestrawhome/rakestrawhome-gemma-4-31b` |
+| Built-in `promptkit-openrouter-standard` | No automatic replacement; optionally define caller profile `openrouter/standard` |
+| Built-in `promptkit-openrouter-auto` | No automatic replacement; optionally define caller profile `openrouter/auto` |
+
+These are conversion examples, not aliases or bundled resources. Mounting old
+`/v3` assets does not update backend references, profile inheritance, prompt
+defaults, or diagnostic defaults inside them. Edit those values to the exact
+qualified IDs you chose.
+
+For example, an optional caller-owned standard profile may extend a selected
+Gemini profile in the same `openrouter` collection:
+
+```yaml
+collections:
+  openrouter:
+    profiles:
+      - id: standard
+        extends: {id: gemini-flash-latest}
+```
+
+An optional auto profile may instead select the exact backend and model values:
+
+```yaml
+collections:
+  openrouter:
+    profiles:
+      - id: auto
+        backend: openrouter/default
+        model: openrouter/auto
+```
+
+Neither optional profile name is a Promptkit contract. Keep provider model
+strings and settings as selected by the application, including any meaningful
+prefixes. If a backend declares diagnostic defaults, update them to exact
+qualified IDs; an unselected diagnostic default is not automatically made
+resolvable by migration.
+
+## Prompt Selection API
+
+Prompt selection is ID-only. Remove version selectors from calls to `Run`,
+`Prepare`, `PrepareExecution`, `ValidateRequest`, `InspectRequirements`, and
+`SmokeTest`. Remove the `PromptVersion` field from `RunRequest`,
+`RequirementInspectionRequest`, and `SmokeTestRequest`; `InspectPrompt` now
+has the form `InspectPrompt(ctx, promptID)`. For example, select the prompt
+explicitly in the request rather than setting a version selector:
+
+```go
+request := promptkit.RunRequest{
+	PromptID:  "support/report",
+	ProfileID: "support/standard",
+}
+_, err = engine.Run(ctx, request)
+if err != nil {
+	return err
+}
+```
+
+If the former format contained multiple versions under one ID, choose the one
+definition the application should install. Promptkit no longer retains or
+returns its former descriptive label.
+
+The public inventory APIs can help an operator inspect a captured effective
+configuration: `ListProfiles`, `ListPrompts`, and `ListBackends` return exact
+identities, while `ListNamespaces` and `ListCollections` expose resource-bearing
+namespaces and effective collection declarations. Inventory does not select
+which configuration version the application should deploy.
+
+## Deployment Selection Remains With The Application
+
+Applications decide which checked-in, embedded, operator-supplied, or
+programmatic inputs to compose for each deployment. They also decide when to
+construct a replacement engine after changing those sources. Promptkit does
+not fetch catalog updates, select provider availability, validate upstream
+catalog release status, or automatically migrate arbitrary consumer files.
+The maintained [offline collection example](../../examples/go-library/prepare/main.go)
+shows explicit source selection, local imports, layering, a scoped view, and
+operator inventory without a provider call.

docs/consumers/pkg-promptkit.md

@@ -12,63 +12,356 @@ file contracts.
 Import the package as:
 
 ```go
-import "gitea.maximumdirect.net/eric/promptkit"
+import "go.maximumdirect.net/promptkit"
 ```
 
+For migration from the former module path, see
+[Update Go Imports](configuration-migration.md#update-go-imports).
+
+Configuration APIs use `go.maximumdirect.net/configkit` v1.1.0. Its types are
+not interchangeable with the old ConfigKit module path.
+
+The maintained [offline preparation example](../../examples/go-library/prepare/main.go)
+demonstrates explicit collection composition and inventory. Applications
+adapting older provider or versioned-prompt configuration can use the
+[configuration migration guide](configuration-migration.md) for the required
+caller-owned choices.
+
 The following Go fragments are illustrative and omit surrounding package,
 import, and error-handling code. Use the maintained examples for complete
 programs.
 
+## Go Compatibility
+
+The `go` directive in the selected release's [module metadata](../../go.mod)
+declares its minimum supported consumer Go version, currently Go 1.27.0.
+Maintainers build and
+validate releases with the locally selected Go patch release; the minimum
+does not simply track that patch version.
+
+Successful release validation on the selected patch is sufficient to presume
+support across its Go series, subject to the declared minimum and concrete
+evidence of incompatibility. Specific compatibility concerns are investigated
+as they arise. Minor releases may raise the minimum; patch releases preserve
+it. Release notes explain increases and any required consumer action. See the
+[release policy](../policy/release.md#consumer-minimum-and-release-toolchain).
+
 ## Construct An Engine
 
-Create an engine with
-[`NewEngine`](../../engine.go). A directory-backed setup supplies a prompt
-directory and may supply profile and schema directories:
+Create an engine with [`NewEngine`](../../engine.go). Select each directory
+explicitly and place its inputs in an ordered ConfigKit underlay. A group
+strictly composes its files; later groups can overlay earlier ones:
+
+The bundled OpenRouter backend and profiles are installed in a lower underlay
+by default. Select a profile explicitly, for example:
 
 ```go
+input := configkit.Input{Source: configkit.Mapping("application", map[string]any{
+	"prompts": []any{map[string]any{
+		"id": "my-prompt",
+		"messages": []any{map[string]any{"role": "user", "content": "Hello"}},
+		"output": map[string]any{"format": "text", "validation_mode": "none"},
+	}},
+})}
+engine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{Main: &input}})
+if err != nil { return err }
+result, err := engine.Run(ctx, promptkit.RunRequest{PromptID: "my-prompt", ProfileID: "gpt-luna-latest"})
+```
+
+The backend looks up
+`OPENROUTER_API_KEY` when available; credentials are not required at engine
+construction. For the complete inventory and diagnostic defaults see the
+[format reference](../formats.md#bundled-openrouter-configuration). Use
+`promptkit.WithBundledOpenRouter(false)` for application-only inventories.
+Caller records replace matching bundled records in full; use `extends` with a
+distinct ID to inherit. Replacing `auto` also changes its cost-tier children's
+inherited target, while their plugins remain. External collections remain
+supported.
+
+```go
+promptInputs, err := promptkit.DirectoryInputs("prompts")
+if err != nil { return err }
+profileInputs, err := promptkit.DirectoryInputs("profiles")
+if err != nil { return err }
 engine, err := promptkit.NewEngine(promptkit.Config{
-	PromptDir:  "prompts",
-	ProfileDir: "profiles",
-	SchemaDir:  "schemas",
+	Sources: promptkit.ConfigurationSources{Underlays: []configkit.Underlay{
+		{Inputs: promptInputs},
+		{Inputs: profileInputs},
+	}},
+	SchemaDir: "schemas",
 })
 ```
 
-Options support single-file or `fs.FS` sources, in-memory profiles,
-engine-scoped backends, and injected artifact or model clients. Consult the
-[constructor and option GoDoc](../../engine.go) for composition, precedence,
-validation, and default transport behavior. Source discovery, format
-validation, and profile precedence are defined by the
-[framework format reference](../formats.md).
+For a single highest-priority document, set `Sources.Main` to a
+`*configkit.Input`; omit it for ordinary layered construction. A single file
+can use `configkit.File` directly. `ProfileSource`, `BackendSource`, and raw
+`configkit.Mapping` supply programmatic definitions. Schema and runtime
+collaborator options remain available. Construction validates the final
+effective configuration and captures referenced prompt content; construct a
+new engine to observe configuration edits. See the [constructor GoDoc](../../engine.go)
+for the exact public contract and the [framework format reference](../formats.md)
+for definition fields.
+
+### Use Scoped Engine Views
+
+Create a view when an application wants short request IDs to resolve within a
+selected collection:
+
+```go
+scoped, err := engine.WithNamespaces(promptkit.Namespaces{
+	Prompt:  "support",
+	Profile: "support",
+})
+if err != nil { return err }
+
+prepared, err := scoped.Prepare(ctx, promptkit.RunRequest{
+	PromptID:  "answer",
+	ProfileID: "concise",
+})
+```
+
+Execution, preparation, prompt/profile inspection, requirement inspection, and
+smoke diagnostics qualify slashless caller selections with their respective
+namespaces. Already-qualified IDs pass through unchanged; configured defaults
+and references are not rewritten. Each call replaces the view's scope defaults,
+including clearing a scope with an empty field, without changing the receiver
+or rebuilding its captured configuration and execution collaborators.
+Namespace syntax is checked when creating the view; resource existence is
+checked when selected. Configuration inspection remains global. See
+[`WithNamespaces`](../../namespaces.go) for the exact public contract.
 
 ## Supply Embedded Application Defaults
 
-Use `WithFallbackProfileFS` when an application packages profile definitions
-that should apply unless an operator provides an ordinary configured profile
-with the same ID. For example, an application can embed its defaults while
-continuing to use `ProfileDir` for operator overrides:
+Use `FilesystemInputs` for embedded defaults and `DirectoryInputs` for an
+explicitly chosen operator directory. A later programmatic group can override
+both. No directory name has built-in precedence:
 
 ```go
-//go:embed profiles/*.yaml
-var applicationProfiles embed.FS
-
-engine, err := promptkit.NewEngine(promptkit.Config{
-	PromptDir:  "prompts",
-	ProfileDir: operatorProfileDir,
-},
-	promptkit.WithFallbackProfileFS(applicationProfiles, "profiles"),
-)
+defaults, err := promptkit.FilesystemInputs("embedded defaults", applicationFS, "config")
+if err != nil { return err }
+operator, err := promptkit.DirectoryInputs(operatorDirectory)
+if err != nil { return err }
+profile := promptkit.ProfileSource("application choices", []promptkit.ProfileConfig{
+	{ID: "summary", Endpoint: &selectedEndpoint, Model: &selectedModel},
+})
+engine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{
+	Underlays: []configkit.Underlay{
+		{Inputs: defaults},
+		{Inputs: operator},
+		{Inputs: []configkit.Input{{Source: profile}}},
+	},
+}})
 ```
 
-Keep application-owned profile IDs and definitions in the embedded source.
-Use the ordinary configured profile source for operator overrides. Leave
-`operatorProfileDir` empty when the operator did not configure an override
-directory; a non-empty path names an authoritative higher-precedence source,
-so an unavailable or unreadable directory is an error rather than a reason to
-fall back. The
-[framework format reference](../formats.md#source-and-profile-precedence)
-owns the exact profile format and lookup order; the
-[`WithFallbackProfileFS` GoDoc](../../engine.go) owns its option contract and
-validation rules.
+Each later same-ID resource replaces the complete earlier record; use a distinct
+inherited sibling for field-level customization. Only add the operator group
+when the application selects that directory; an
+explicitly selected missing or unreadable root is an error. Each helper finds
+`.yaml` and `.yml` files recursively and disables imports by default. The
+caller may enable imports on individual inputs. A nil `Sources.Main` is an
+empty main source, not another directory lookup.
+
+For a programmatic document with explicit null or a prompt definition, use a
+labeled mapping as an input. A mapping owns a snapshot of supported values:
+
+```go
+document := configkit.Mapping("application configuration", map[string]any{
+	"profiles": []any{
+		map[string]any{"id": "local-summary",
+			"backend": "local", "model": "example-model",
+			"temperature": nil, // Suppress an inherited provider control.
+		},
+	},
+})
+input := configkit.Input{Source: document}
+// Add input to a later underlay, or set Sources.Main = &input.
+```
+
+`ProfileSource` and `BackendSource` are smaller typed conveniences when null
+is not needed. Backend and profile definitions are supplied by the caller;
+there is no registration-only backend namespace.
+
+When configuration or message files change, build and validate a new engine,
+then switch the application to that engine for subsequent work. Existing
+engines and prepared handles retain their captured definitions; schema and
+request artifact resources still follow their runtime lifecycle. Inspection
+does not watch or reload sources automatically.
+
+## Discover Available Resources And Collections
+
+Use [`Engine.ListProfiles`](../../listing.go),
+[`Engine.ListPrompts`](../../listing.go), and
+[`Engine.ListBackends`](../../listing.go) to discover exact identities from the
+engine's final captured configuration. A scoped view filters by its respective
+namespace by default; pass `NamespaceNamed`, `NamespaceUnqualified`, or
+`NamespaceAll` to override that listing filter without changing execution
+scope. Returned IDs remain full identities and can be used for exact selection:
+
+```go
+profiles, err := engine.ListProfiles(ctx)
+if err != nil {
+	return err
+}
+if len(profiles) > 0 {
+	_, err := engine.InspectProfile(ctx, profiles[0])
+	if err != nil {
+		return err
+	}
+}
+prompts, err := engine.ListPrompts(ctx)
+if err != nil {
+	return err
+}
+if len(prompts) > 0 {
+	selected := prompts[0]
+	inspection, err := engine.InspectPrompt(ctx, selected.ID)
+	if err != nil {
+		return err
+	}
+	_ = inspection
+}
+```
+
+Use [`Engine.ListNamespaces`](../../listing.go) to discover resource-bearing
+namespaces across the entire captured configuration, independent of view
+scopes. [`Engine.ListCollections`](../../listing.go) reports effective
+collection declarations, including metadata-only entries; it does not infer
+collection metadata from resource membership or source history.
+
+These lists describe effective membership, not ConfigKit contribution history,
+and do not promise that a profile is ready to execute. Use inspection for
+details and preparation to resolve request resources. The linked method GoDoc
+owns ordering, ownership, concurrency, and error behavior; identifiers may be
+sensitive application data.
+
+## Inspect The Captured Configuration
+
+Use [`Engine.InspectConfiguration`](../../configuration_inspection.go) for an
+operator-selected report of resolved configuration, source descriptions,
+diagnostics, and path history. Configuration inspection uses list indexes in
+the collection's nested path rather than qualified runtime IDs. For a profile
+at index `i` in collection `team`, construct the path with `configkit.Keys`:
+
+```go
+view, err := engine.InspectConfiguration()
+if err != nil { return err }
+path := configkit.Keys("collections", "team", "profiles").Index(i).Key("model")
+explanation, found := view.Explain(path)
+if found && explanation.Effective.Redacted {
+	// Apply the application's disclosure policy before sharing the report.
+}
+```
+
+`WriteEffective` and `WriteExplanation` produce ConfigKit-redacted reports;
+the application decides whether those reports may be disclosed. Credential-
+shaped provider keys are redacted, but arbitrary prompt and provider content
+is not necessarily safe to share. Inspection uses the engine's captured result
+and does not reopen sources or resolve credentials. It retains effective values,
+including recorded field normalization and queue defaulting, alongside source
+provenance.
+`InspectProfile` instead shows backend-completed execution settings for one profile, and `Prepare` or
+`PreparedExecution.Details` shows request-specific rendered messages, inputs,
+and validation resources. None of those views is a substitute for the others.
+
+To review changes between two independently captured configurations, compare
+their inspection views:
+
+```go
+earlier, err := earlierEngine.InspectConfiguration()
+if err != nil {
+	return err
+}
+later, err := candidateEngine.InspectConfiguration()
+if err != nil {
+	return err
+}
+differences, err := earlier.Diff(later)
+if err != nil {
+	return err
+}
+for _, difference := range differences {
+	if difference.HiddenChanged {
+		// Report a protected change without inferring its value.
+	}
+	if difference.Left.Redacted || difference.Right.Redacted {
+		// Apply the application's disclosure policy before presenting snapshots.
+	}
+}
+```
+
+See the [`ConfigurationInspection.Diff` GoDoc](../../configuration_inspection.go)
+and ConfigKit's [v1.1.0 effective-configuration diff guidance](https://forge.maximumdirect.net/base-layer/configkit/src/tag/v1.1.0/docs/config.md#effective-configuration-diffs)
+for the exact comparison contract. The operation compares only frozen effective
+configuration values: it does not reread external files or compare their
+contents, resolved credential contents, later execution preparation, or
+request-specific overrides. Inputs normalizing to the same effective value do
+not create a value diff; lists are compared atomically. An empty diff is not
+proof of equivalent execution.
+Records can contain ordinary application data, so use the application's
+disclosure policy when presenting them.
+
+## Run Explicit Smoke Diagnostics
+
+[`Engine.SmokeTest`](../../smoke_diagnostics.go) runs one explicit diagnostic
+using a target resolved from the engine's captured configuration. It does not
+execute the selected application prompt or its inputs, session, artifacts, or
+output schema. Unlike the offline examples and default test suite, calling it
+can contact the selected provider and use configured credentials. It is a
+consumer operation, not an automatic readiness verdict or a release check.
+
+For example, a caller can request the built-in text inference check and treat
+its optional fact as an observation rather than assuming that a nil Go error
+means success:
+
+```go
+report, err := engine.SmokeTest(ctx, promptkit.SmokeTestRequest{
+	Kind:      promptkit.SmokeTestTextInference,
+	ProfileID: profileID,
+	APIKey:    apiKey,
+})
+if err != nil {
+	if report != nil {
+		retainDisclosureSafeSummary(report)
+	}
+	return err
+}
+if report != nil {
+	matched := report.Facts.ExpectedTextMatched
+	switch {
+	case matched == nil:
+		// The comparison was not evaluated or is unknown/inapplicable.
+	case *matched:
+		// The private expected-text comparison matched.
+	default:
+		// It did not match; the operation may still have concluded normally.
+	}
+}
+```
+
+A nil Go error means the diagnostic concluded, not that its finding was
+positive. Fact pointers are tri-state: nil means unknown, not evaluated, or
+inapplicable. Reports contain factual observations and effective settings, not
+generated text or provider bodies; identifiers and settings can still be
+sensitive. The
+[`SmokeTest` GoDoc](../../smoke_diagnostics.go) owns selection precedence,
+exact fields, timeout and retry behavior, supported transports, and disclosure
+details.
+
+Caller cancellation can return both an error and a partial report. Preserve the
+context error identity and inspect any available observations before returning:
+
+```go
+report, err := engine.SmokeTest(ctx, request)
+if report != nil {
+	retainDisclosureSafeSummary(report)
+}
+if err != nil {
+	if errors.Is(err, context.Canceled) || errors.Is(err, context.DeadlineExceeded) {
+		// The report may describe work completed before interruption.
+	}
+	return err
+}
+```
 
 ## Inspect A Prompt Before Preparation
 
@@ -77,7 +370,7 @@ declared inputs and output workflow without creating placeholder inputs or
 resolving a profile:
 
 ```go
-inspection, err := engine.InspectPrompt(ctx, "meeting.summary", "")
+inspection, err := engine.InspectPrompt(ctx, "meeting.summary")
 if err != nil {
 	return err
 }
@@ -94,6 +387,85 @@ rendered messages, and use prepared execution when that work must remain tied
 to later execution. The method's [GoDoc](../../engine.go) owns exact fields,
 hash, ownership, and error semantics.
 
+## Inspect Requirements Before Preparation
+
+Use [`Engine.InspectRequirements`](../../requirements.go) when the application
+needs the prompt's recognizable input and caller-variable dependencies for a
+particular context before it has input artifacts. Unlike `InspectPrompt`, it
+also walks captured template syntax and selects a profile identity; unlike
+`Prepare`, it does not load artifacts or render messages.
+
+```go
+requirements, err := engine.InspectRequirements(ctx, promptkit.RequirementInspectionRequest{
+	PromptID: "meeting.summary",
+	Vars:     map[string]string{"audience": "project leads"},
+})
+if err != nil {
+	return err
+}
+
+for _, input := range requirements.Inputs {
+	if input.Kind == promptkit.RequirementRequired {
+		// Plan to bind this input before preparation.
+	}
+	if input.Kind == promptkit.RequirementOptional && input.Explanation != "" {
+		// A recognizable reference may occur only on a known skipped path.
+	}
+}
+for _, variable := range requirements.Variables {
+	if variable.Kind == promptkit.RequirementRequired && !variable.Supplied {
+		// The key is required even though this inspection did not receive it.
+	}
+}
+if !requirements.Complete {
+	// Preserve the partial inventory and account for its discovery issues.
+}
+```
+
+Optional entries can include recognizable inputs and variables found only in
+inactive branches. A missing required variable is reported separately from
+discovery completeness: missing a value used only for direct text substitution
+does not by itself make the inventory incomplete. Conversely, an unresolved
+dependency-bearing branch can make discovery incomplete while retaining known
+conditional entries. Do not treat an empty partial inventory as proof that no
+input bindings are needed. See the method's [GoDoc](../../requirements.go) for
+the exact classifications, supported syntax, ownership, and error contract.
+
+## Validate A Concrete Request
+
+Use [`Engine.ValidateRequest`](../../engine.go) when the application has a
+complete `RunRequest` and needs to know whether that concrete invocation passes
+preparation:
+
+```go
+request := promptkit.RunRequest{
+	PromptID: "meeting.summary",
+	Vars:     map[string]string{"audience": "project leads"},
+	Inputs: map[string]promptkit.ArtifactRef{
+		"note": promptkit.Inline("Synthetic meeting notes"),
+	},
+}
+if err := engine.ValidateRequest(ctx, request); err != nil {
+	return err
+}
+```
+
+Unlike `InspectRequirements`, this checks supplied values, the selected template
+branches, and request bindings through the existing preparation path. It checks
+only this invocation, does not aggregate errors, and loads every supplied input
+according to ordinary preparation behavior, even if a branch does not use it.
+It makes no model call and returns no prepared snapshot, but artifact and schema
+resource I/O and its time and memory costs remain. Success does not certify
+provider acceptance or later execution success. See the method's GoDoc for its
+exact error and context behavior.
+
+Scriptorium can map the returned error to its own diagnostics and exit status.
+Its `validate` and `run` commands should construct equivalent requests for
+equivalent caller inputs so their preparation behavior agrees. Separate
+subprocess calls prepare independently; an opaque `PrepareExecution` handle
+cannot be transferred from `validate` to `run`. Use `PrepareExecution` when one
+process and engine must execute the exact prepared snapshot.
+
 ## Prepare Without Model Execution
 
 [`Engine.Prepare`](../../engine.go) resolves the selected prompt and profile,
@@ -114,8 +486,11 @@ prepared, err := engine.Prepare(ctx, promptkit.RunRequest{
 The maintained
 [offline preparation example](../../examples/go-library/prepare/main.go)
 shows a complete runnable setup with a prompt file, in-memory profile, and
-inline input. Exact request requirements and prepared-result fields belong to
-the [`RunRequest` and `PreparedRun` GoDoc](../../types.go).
+inline input. It demonstrates validation and preparation with the same request
+to make the two operations concrete; this performs preparation twice and is
+illustrative, not a requirement to validate before `Prepare` or `Run`. Exact
+request requirements and prepared-result fields belong to the
+[`RunRequest` and `PreparedRun` GoDoc](../../types.go).
 
 ## Prepare Now And Execute The Same Snapshot Later
 
@@ -151,6 +526,14 @@ handling policy. The
 [engine operation GoDoc](../../engine.go) own exact lifecycle, engine-binding,
 credential, cancellation, timing, and error semantics.
 
+If the request supplied `Observer`, successful preparation retains that
+function for `RunPrepared`; preparation itself emits nothing. Observation
+starts only when the owning engine successfully claims the handle. Reused,
+discarded, foreign, and invalid handles emit nothing. `Discard` clears the
+retained callback without invoking it. The function's closure is retained by
+identity, so synchronize any shared state it uses. Prepared start events can
+include identities already resolved in the frozen snapshot.
+
 ## Execute And Validate
 
 [`Engine.Run`](../../engine.go) performs the same preparation, invokes the
@@ -173,6 +556,98 @@ semantics. The
 [OpenAI-compatible integration contract](../integrations/openai-compatible-chat.md)
 owns the built-in client's outbound HTTP behavior.
 
+### Observe An Ordinary Run
+
+Set `RunRequest.Observer` to receive synchronous metadata events from `Run`:
+
+```go
+request.Observer = func(event promptkit.ExecutionEvent) {
+	// Translate selected metadata into an application-owned record.
+}
+result, err := engine.Run(ctx, request)
+```
+
+An ordinary run emits `started` before request conversion, followed by any
+attempt and retry-delay completions, then one `finished` event on a normal
+return. `Prepare` and `PrepareExecution` emit no events. A start event has no
+resolved prompt, profile, or model; later events gain identities only as
+selection succeeds. The terminal event has dispatched-attempt, transient-retry,
+repair-call, duration, and presence-aware usage totals. Its usage is an
+aggregate, not an additional billable attempt. A known zero differs from an
+unknown count, and incomplete totals must not be treated as complete cost
+accounting. A failed operation still returns no partial `RunResult`.
+An injected client that leaves `UsagePresence` nil can report known nonzero
+counts, but its zero counts remain unknown to observations; supply explicit
+presence when measured zeros matter. Existing `RunResult.Usage` sums returned
+values independently of observation presence.
+
+Callbacks run synchronously, in order for one operation, and can overlap for
+different operations. Synchronize shared callback state and avoid blocking on
+work that needs the same run's admission slot. Callback panics propagate;
+there is no durable event history or promise of a terminal event after panic.
+Events omit content and error objects, but selected identifiers may still be
+sensitive. See the [observation GoDoc](../../observations.go) for exact fields.
+The [offline proposal and assessment example](../../examples/go-library/observe/main.go)
+shows application-owned event collection and a separate semantic acceptance
+decision using the built-in HTTP path without network access.
+
+### Repair A Structured Result
+
+Set a small additional-call budget when a structurally invalid result can be
+corrected automatically:
+
+```go
+request.Validation = &promptkit.OutputContract{
+	Format:         promptkit.FormatJSON,
+	ValidationMode: promptkit.ValidationJSONSchema,
+	SchemaPath:     "events.schema.json",
+	RepairAttempts: 1,
+}
+```
+
+Each repair attempt is another model generation, so it can increase latency and
+usage; `RunResult.Usage` is cumulative and `Validation.RepairAttempts` reports
+repair generations rather than transient HTTP attempts. Exhaustion still
+returns the final failed validation result. `basic` validation can also repair
+an empty candidate, but structural
+validity is not evidence of factual or domain correctness. See the
+[output-contract format reference](../formats.md#output-contract) and
+[`OutputContract` GoDoc](../../types.go) for the exact budget and eligibility
+rules.
+
+The built-in client also retries eligible temporary HTTP statuses. By default,
+each generation can make two additional attempts; a repair generation receives
+its own budget. Set `Config.TransientRetries` to a pointer to zero for the
+previous single-attempt behavior, or set `RunRequest.TransientRetries` to
+override one execution. A custom HTTP transport still uses this policy;
+injected `LLMClient` implementations retain their own retry policy. Application
+semantic retries are separate runs and can multiply provider traffic further.
+The caller context covers capacity and retry waiting as well as attempts;
+without a deadline a valid provider delay can be long. Promptkit does not
+infer exactly-once or complete cost accounting from a retry outcome.
+See the [provider integration contract](../integrations/openai-compatible-chat.md#transient-status-retries)
+for eligible statuses, waits, timeouts, and provider timing.
+
+### Append Already-Rendered Messages
+
+An application can include an earlier assistant response and its own corrective
+instruction in a fresh request without changing the configured prompt:
+
+```go
+request.AppendedMessages = []promptkit.RenderedMessage{
+	{Role: promptkit.RoleAssistant, Content: previousResponse},
+	{Role: promptkit.RoleUser, Content: correction},
+}
+result, err := engine.Run(ctx, request)
+```
+
+These messages are already rendered: Promptkit does not template or resolve
+files in them, and they can contain sensitive model output or application
+feedback. Every `Run` call uses the engine's captured configuration; the
+application owns any semantic retry budget. When a
+pre-execution equality check is required, use `PrepareExecution` and compare
+its opaque rendered-prompt hash before invoking `RunPrepared`.
+
 ## Inputs, Profiles, And Overrides
 
 Use `File`, `Inline`, or `InlineWithURI` to construct request inputs. A request
@@ -182,59 +657,48 @@ replace execution settings or the complete output contract.
 The [public value GoDoc](../../types.go) defines nil, empty, zero, replacement,
 copy, and credential behavior. The
 [framework format reference](../formats.md) defines how those request values
-interact with prompt definitions, file-backed and application fallback
-profiles, built-ins, schemas, and framework defaults.
+interact with prompt definitions, layered profiles, built-ins, schemas, and
+framework defaults.
 
-For programmatic profiles,
-[`OpenAICompatibleProfile`](../../profiles.go) converts ordinary
-OpenAI-compatible settings into a value accepted by `WithProfiles`.
+For programmatic profiles, use [`ProfileSource`](../../profiles.go) with identified
+`ProfileConfig` values. Nil fields are omitted; pointers preserve explicit
+zero, false, and blank values. Raw `configkit.Mapping` expresses explicit null.
 
-### Alias A Built-In Profile
+### Migrate From The Previous Configuration API
 
-Give an application-owned profile ID a built-in base when prompts should select
-the application ID while inheriting the built-in target. The child can override
-only the setting it owns:
+Replace `Config.PromptDir` and `Config.ProfileDir` with selected ConfigKit
+inputs in `Config.Sources`. The removed `WithPromptFS`, `WithPromptFile`,
+`WithProfileFS`, `WithProfileFile`, `WithFallbackProfileFS`, `WithProfiles`,
+and `WithBackend` options have no compatibility wrappers. Use
+`DirectoryInputs` or `FilesystemInputs` for a strict-composition file group,
+ConfigKit file inputs for individual files, and `ProfileSource`,
+`BackendSource`, or `configkit.Mapping` for programmatic values. Backend IDs
+are fields in configuration, not registrations. Choose caller underlays in
+fallback-to-operator-to-programmatic order; use `Sources.Main` for one final
+highest-priority input.
 
-```go
-promptkit.WithProfiles(promptkit.Profile{
-	ID:              "weather-light",
-	BaseProfileID:   "deepseek-4-flash",
-	ReasoningEffort: "high",
-})
-```
+Move resource IDs into identified `backends`, `profiles`, and `prompts` lists;
+each prompt ID identifies one installed definition without a version label.
+Convert `base_profile` to ConfigKit `extends: {id: parent}`. For the current
+collection identity mapping, ID-only prompt selection, and required
+caller-owned choices, see the
+[configuration migration guide](configuration-migration.md). Exact current
+format, reference, inheritance, and presence semantics belong to the
+[framework format reference](../formats.md). Prompt text and static templates
+are captured at construction, not automatically reloaded between runs. Replace
+the engine to pick up changed definitions or message files.
 
-Select `weather-light` in a prompt or `RunRequest.ProfileID`; it remains the
-reported selected profile. See the [profile inheritance format
-reference](../formats.md#profile-inheritance) and the
-[`Profile` GoDoc](../../types.go) for exact lookup, merging, and validation
-behavior.
+### Choose A Profile Ownership Model
 
-### Use The Rakestrawhome Built-In Profile
-
-Set `RAKESTRAWHOME_INFERENCE_API_KEY` in the application environment, then
-select `rakestrawhome-gemma-4-31b` as an ordinary profile ID. For example, a
-prepared result identifies the selected built-in through
-`BackendRakestrawHome`:
-
-```go
-prepared, err := engine.Prepare(ctx, promptkit.RunRequest{
-	PromptID:  "meeting.summary",
-	ProfileID: "rakestrawhome-gemma-4-31b",
-	Inputs:    inputs,
-})
-if err != nil {
-	return err
-}
-if prepared.SelectedBackendID != promptkit.BackendRakestrawHome {
-	return fmt.Errorf("unexpected backend %q", prepared.SelectedBackendID)
-}
-```
-
-Do not register `rakestrawhome` manually. When adopting this built-in, remove
-an existing `WithBackend` registration with that exact ID; retaining it causes
-the intentional duplicate-ID configuration error. Direct request credentials
-and runtime endpoint overrides remain supported under their ordinary GoDoc and
-format contracts.
+Supply backend and profile definitions through the configuration sources your
+application selects, alongside the default-on bundled OpenRouter inventory.
+Conventional backend ID constants are names only. Define or replace the model,
+endpoint, credentials lookup, and settings that your application
+intends to support. A profile may inherit a sibling profile by its local key in
+the same root or collection list; its backend reference uses the exact
+configured identity. Caller-selected source layers determine the final values. The
+[format reference](../formats.md) describes the definition fields and
+composition rules.
 
 ### Inspect A Profile Before Prompt Work
 
@@ -299,19 +763,15 @@ Choose the smallest configuration that fits how the endpoint will be reused.
 
 #### Use An Endpoint-Only Profile
 
-Put the endpoint directly on an in-memory profile when only that profile needs
+Put the endpoint directly on a programmatic profile when only that profile needs
 it and shared backend identity or capacity policy is unnecessary:
 
 ```go
-engine, err := promptkit.NewEngine(promptkit.Config{
-	PromptDir: "prompts",
-},
-	promptkit.WithProfiles(promptkit.Profile{
-		ID:       "local-summary",
-		Endpoint: "http://localhost:8000/v1",
-		Model:    "example-model",
-	}),
-)
+endpoint, model := "http://localhost:8000/v1", "example-model"
+profile := promptkit.ProfileSource("local profile", []promptkit.ProfileConfig{
+	{ID: "local-summary", Endpoint: &endpoint, Model: &model},
+})
+// Add configkit.Input{Source: profile} to a caller underlay.
 ```
 
 Endpoint-only profiles have an empty backend ID and remain unrestricted by
@@ -323,59 +783,47 @@ Use `LocalBackend` when profiles should share the conventional `local`
 identity, endpoint, and concurrency limit:
 
 ```go
-engine, err := promptkit.NewEngine(promptkit.Config{
-	PromptDir: "prompts",
-},
-	promptkit.WithBackend(
-		promptkit.LocalBackend("http://localhost:8000/v1", 2),
-	),
-	promptkit.WithProfiles(promptkit.Profile{
-		ID:        "local-summary",
-		BackendID: promptkit.BackendLocal,
-		Model:     "example-model",
-	}),
-)
+backendID, model := promptkit.BackendLocal, "example-model"
+localBackend := promptkit.LocalBackend("http://localhost:8000/v1", 2)
+localBackend.ID = backendID
+backend := promptkit.BackendSource("local backend", []promptkit.BackendConfig{
+	localBackend,
+})
+profile := promptkit.ProfileSource("local profile", []promptkit.ProfileConfig{
+	{ID: "local-summary", Backend: &backendID, Model: &model},
+})
+// Add both sources as inputs in a caller underlay.
 ```
 
-The helper is explicit: it does not pre-register a backend or read environment
-variables. Supplying a positive limit leaves queue capacity omitted, so normal
-backend registration selects the existing default waiting capacity of 1024.
-The returned value still enters the engine through `WithBackend`.
+The helper does not choose an ID or read environment variables. A positive
+concurrency limit with omitted queue capacity selects the default waiting
+capacity of 1024.
 
 #### Configure A Complete Backend
 
-Use a keyed `Backend` value for authentication, extra request parameters, an
+Use an identified `BackendConfig` value for authentication, extra request parameters, an
 explicit queue capacity, a custom ID, or multiple local endpoints:
 
 ```go
-noWaiting := 0
-engine, err := promptkit.NewEngine(promptkit.Config{
-	PromptDir: "prompts",
-},
-	promptkit.WithBackend(promptkit.Backend{
-		ID:               "local-gpu",
-		Endpoint:         "http://gpu-host:8000/v1",
-		APIKeyEnv:        "LOCAL_GPU_API_KEY",
-		ExtraParams:      map[string]any{"provider_option": "enabled"},
-		ConcurrencyLimit: 2,
-		QueueCapacity:    &noWaiting,
-	}),
-	promptkit.WithProfiles(promptkit.Profile{
-		ID:        "gpu-summary",
-		BackendID: "local-gpu",
-		Model:     "example-model",
-	}),
-)
+endpoint, keyName, limit, noWaiting := "http://gpu-host:8000/v1", "LOCAL_GPU_API_KEY", 2, 0
+backend := promptkit.BackendSource("GPU backend", []promptkit.BackendConfig{
+	{
+		ID: "local-gpu",
+		Endpoint: &endpoint, APIKeyEnv: &keyName,
+		ConcurrencyLimit: &limit, QueueCapacity: &noWaiting,
+		ExtraParams: map[string]any{"provider_option": "enabled"},
+	},
+})
 ```
 
-Use distinct custom IDs when registering multiple local endpoints.
-Registrations belong to one engine and custom IDs cannot replace built-ins.
-The [`Backend`, `LocalBackend`, and `WithBackend` GoDoc](../../backends.go)
-defines exact construction, validation, copying, uniqueness, concurrency, and
-request-default behavior.
+Place `backend` in a caller underlay and select `"local-gpu"` from a
+`ProfileConfig.Backend` pointer or an identified file profile. All backend
+contributions use the same ConfigKit layering rules, including overlays of
+maintained definitions. The [`BackendConfig` and `LocalBackend` GoDoc](../../backends.go)
+defines exact omission and construction behavior.
 
-Both file-backed and in-memory profiles select a registration through
-`backend` or `Profile.BackendID`. Profile and request endpoint overrides retain
+Both file-backed and programmatic profiles select a backend through
+`backend` or `ProfileConfig.Backend`. Profile and request endpoint overrides retain
 that routing and capacity identity. `PreparedRun.SelectedBackendID`,
 `RunResult.SelectedBackendID`, and the effective `ExecutionTarget.BackendID`
 expose it to consumers and injected model clients. Endpoint-only profiles
@@ -383,7 +831,7 @@ remain supported and expose an empty backend ID.
 
 ### Limit Backend Concurrency
 
-Set `Backend.ConcurrencyLimit` when a backend needs protection from too many
+Set `BackendConfig.ConcurrencyLimit` when a backend needs protection from too many
 simultaneous model calls. Leaving `QueueCapacity` nil, as in the local-backend
 example above, selects the default waiting capacity of 1024.
 
@@ -392,24 +840,23 @@ zero:
 
 ```go
 noWaiting := 0
-backend := promptkit.Backend{
-	ID:               "local-gpu",
-	Endpoint:         "http://gpu-host:8000/v1",
-	ConcurrencyLimit: 2,
-	QueueCapacity:    &noWaiting,
+backend := promptkit.BackendConfig{
+	Endpoint: &endpoint,
+	ConcurrencyLimit: &limit,
+	QueueCapacity: &noWaiting,
 }
 ```
 
-The pointer distinguishes an explicit zero from omission. Keep using keyed
-`Backend` literals so additive configuration fields remain source-compatible.
+The pointer distinguishes an explicit zero from omission. Use identified
+`BackendConfig` literals inside `BackendSource` for programmatic definitions.
 Capacity belongs to one engine and the selected backend ID; endpoint-only
 profiles and custom backends without a configured limit remain unrestricted.
 Exact validation, defaulting, ownership, and concurrency semantics belong to
-the [`Backend` GoDoc](../../backends.go).
+the [`BackendConfig` GoDoc](../../backends.go).
 
 ## Credentials
 
-File-backed profiles name an environment variable; in-memory profiles can
+File-backed and programmatic profiles can name an environment variable or
 require a direct request key. Direct keys are request-scoped and are excluded
 from supported JSON values and the package's `String` and `GoString`
 summaries. The exact precedence and redaction guarantees belong to
@@ -443,12 +890,13 @@ the application policy that motivated the injection.
 
 Use `errors.Is` with the
 [public error sentinels and operation GoDoc](../../engine.go). The declarations
-distinguish invalid construction, invalid requests, absent sources,
-source-loading failures, collaborator failures, and operational validation
+distinguish invalid construction, invalid requests, absent selections,
+collaborator failures, and operational validation
 failures. Specific request conditions may also match the broader
 `ErrInvalidRequest`, and injected collaborator identities are preserved where
-documented. Invalid or duplicate backend registrations match
-`ErrInvalidConfig`; selecting an unknown backend matches `ErrProfileLoad`.
+documented. Invalid effective definitions, including unknown backend
+references, match `ErrInvalidConfig` during construction. ConfigKit load
+errors remain available through `errors.As`.
 
 When a limited backend has admitted all active and waiting calls, handle
 `ErrCapacityExceeded` separately from request errors and provider failures:
@@ -480,16 +928,30 @@ var generationErr *promptkit.GenerationError
 if errors.As(err, &generationErr) {
 	status := generationErr.StatusCode()
 	message := generationErr.ProviderMessage()
-	_, _ = status, message // Apply application retry and presentation policy.
+	_, _ = status, message // Apply application presentation policy.
 }
 ```
 
 All provider fields are untrusted and can contain sensitive request or schema
 fragments. Do not log, display, or return them without an application-specific
-disclosure policy. Promptkit does not assign retry or presentation behavior.
+disclosure policy. Promptkit does not assign presentation behavior.
 The [`GenerationError` GoDoc](../../generation_error.go) owns the exact typed
 error contract.
 
+Generation failures also match `ErrTransport` when the built-in HTTP request
+fails, or `ErrMalformedResponse` when a successful provider response cannot be
+decoded or an injected client returns a nil successful response. Both retain
+`ErrLLMGenerate`; transport failures retain their original causes for
+`errors.Is` and `errors.As`. Caller cancellation and deadlines retain their
+standard context identities. Arbitrary injected-client errors retain their own
+identities without Promptkit guessing a failure category from their text.
+An eligible built-in status that consumes a positive retry budget additionally
+matches `ErrRetriesExhausted`; use `errors.As` to read
+`RetryExhaustedError.Attempts()` and the final `GenerationError` status. With
+explicit zero retries, the first eligible status is an ordinary generation
+error, not exhaustion. Completed structural rejection remains a normal result
+with failed validation, while an operational error returns no partial result.
+
 ## Application Boundary
 
 Promptkit is an importable library. It does not own a command, inbound HTTP

docs/development.md

@@ -1,8 +1,7 @@
 # Development
 
-This is the contributor entry point for Promptkit, a reusable Go library. All
-contributors must read the
-[architecture policy](policy/architecture.md) before making changes.
+This is the contributor entry point for Promptkit, a reusable Go library.
+Use the Policy Reading Guide below to select policies applicable to the task.
 
 ## Initial Orientation
 
@@ -15,17 +14,27 @@ Before starting work:
 4. keep documentation limited to implemented behavior unless an accepted
    decision or temporary roadmap explicitly owns future work.
 
-Start with:
+## Policy Reading Guide
 
-- the [architecture policy](policy/architecture.md) for library boundaries,
-  dependency direction, invariants, and non-goals;
-- the [internal component overview](internal/overview.md) for the current
-  package and component inventory;
-- the [documentation policy](policy/documentation.md) before changing
-  documentation;
-- the [testing policy](policy/testing.md) before adding, rewriting, or deleting
-  tests; and
-- the [release procedure](release.md) for version and publication work.
+| Policy | Read before… |
+| --- | --- |
+| [Policy set provenance](policy/meta.md) | Adopting policies, checking upstream revisions, or updating the adopted policy set. |
+| [Agents](policy/agents.md) | Recommending, planning, or implementing repository changes. |
+| [Architecture](policy/architecture.md) | Planning or changing library code or its structure. |
+| [Testing](policy/testing.md) | Selecting validation for code changes, or adding, changing, removing, or evaluating tests. |
+| [Documentation](policy/documentation.md) | Writing, revising, or reorganizing documentation. |
+| [OSS](policy/oss.md) | Adding or updating dependencies or incorporated third-party material, or changing licensing or attribution. |
+| [Release](policy/release.md) | Preparing or publishing a release, changing release procedures, or making versioning and compatibility decisions. |
+
+- Read applicable policies before making decisions they govern; multiple policies
+  may apply.
+- Select policies using the task triggers above. Follow cross-references to
+  relevant sections when needed to resolve a concrete question.
+- Reuse policies already available in context; reread when changed or no longer
+  available.
+- Expand reading when the task's scope expands.
+- When revising a policy itself, read that policy and any provisions elsewhere
+  affected by the revision.
 
 ## Task-Specific Reading Guide
 
@@ -34,12 +43,14 @@ Start with:
 | Root public API | The [architecture policy](policy/architecture.md), [consumer guide](consumers/pkg-promptkit.md), [testing policy](policy/testing.md), and existing GoDoc. |
 | Prompt, profile, or schema formats | The [framework format reference](formats.md), owning parser or validator package, and [documentation policy](policy/documentation.md). |
 | Source loading or validation | The [framework format reference](formats.md), [internal source document](internal/sources.md), and owning package tests. |
+| Caller-supplied configuration sources | The [internal source document](internal/sources.md), [framework format reference](formats.md), and [testing policy](policy/testing.md). |
 | Model-client behavior | The [OpenAI-compatible integration contract](integrations/openai-compatible-chat.md), [internal model-client document](internal/llm.md), and owning package tests. |
 | Internal package implementation | The [architecture policy](policy/architecture.md), [internal component overview](internal/overview.md), and focused internal document listed for that package. |
 | Tests or test fixtures | The [testing policy](policy/testing.md), owning package, and focused internal document listed by the component overview. |
-| Maintained example | The [example](../examples/go-library/prepare/main.go), [consumer guide](consumers/pkg-promptkit.md), [framework format reference](formats.md), and [documentation policy](policy/documentation.md). |
+| Maintained example | The [preparation](../examples/go-library/prepare/main.go), [execution](../examples/go-library/run/main.go), and [observation](../examples/go-library/observe/main.go) examples, [consumer guide](consumers/pkg-promptkit.md), [framework format reference](formats.md), and [documentation policy](policy/documentation.md). |
 | Documentation | The [documentation policy](policy/documentation.md) and canonical owner of every affected contract. |
-| Release preparation or publication | The [release procedure](release.md). |
+| Licensing, dependencies, copied material, or attribution | The [OSS policy](policy/oss.md), [project grant](../COPYRIGHT), [license texts](../LICENSE), and [attribution entries](oss/attribution.md); inspect the selected upstream source's notices. |
+| Release preparation or publication, including release notes | The [release policy](policy/release.md), [release procedure](release.md), and [OSS discoverability rules](policy/oss.md#discoverability). |
 
 For cross-cutting changes, follow every applicable row. Do not create
 placeholder documents for packages, APIs, or integrations that do not yet
@@ -47,187 +58,45 @@ exist.
 
 ## Maintainer Validation
 
-This section is the canonical local validation workflow for Promptkit. Run
-every command from the repository root before accepting a change. The test
-suite and maintained examples are deterministic, offline, and require no real
-provider credentials.
-
-### Tests, Analysis, Build, And Examples
-
-Run the ordinary and race-enabled suites, static analysis, the build, and both
-maintained consumer examples:
+For ordinary work, select validation proportionate to the change. Do not repeat
+successful checks for unchanged content merely to create a commit. During
+development, run relevant Go tests and vet, format changed Go files, and execute
+the maintained examples when they are affected:
 
 ```sh
 go test ./...
-go test -race ./...
 go vet ./...
-go build ./...
-go run ./examples/go-library/prepare
-go run ./examples/go-library/run
-```
-
-Both examples must exit successfully. Review their JSON output: preparation
-must report the selected offline prompt, profile, model, and message count;
-execution must report the deterministic generated output, successful
-validation, selected offline model, and usage. Neither command may contact a
-provider or require credentials.
-
-### Go Formatting
-
-Check every tracked Go file. The final command must succeed and the captured
-list must be empty:
-
-```sh
-unformatted=$(
-    git ls-files '*.go' |
-        while IFS= read -r go_file
-        do
-            gofmt -l "$go_file"
-        done
-)
-test -z "$unformatted"
-```
-
-### Local Markdown Links
-
-Use the Python standard library to verify every repository-relative Markdown
-target and local heading fragment. The check is offline and prints nothing on
-success:
-
-```sh
-python3 - <<'PY'
-from pathlib import Path
-import re
-import subprocess
-import sys
-from urllib.parse import unquote
-
-root = Path.cwd().resolve()
-markdown_files = [
-    root / name
-    for name in subprocess.check_output(
-        ["git", "ls-files", "*.md"], text=True
-    ).splitlines()
-]
-link_pattern = re.compile(r"!?\[[^]]*\]\(([^)]+)\)")
-heading_pattern = re.compile(r"^#{1,6}\s+(.+?)\s*#*\s*$")
-scheme_pattern = re.compile(r"^[a-z][a-z0-9+.-]*:", re.IGNORECASE)
-
-
-def markdown_lines(path):
-    in_fence = False
-    fence = ""
-    for line in path.read_text(encoding="utf-8").splitlines():
-        stripped = line.lstrip()
-        marker = stripped[:3]
-        if marker in {"```", "~~~"}:
-            if not in_fence:
-                in_fence = True
-                fence = marker
-            elif marker == fence:
-                in_fence = False
-                fence = ""
-            continue
-        if not in_fence:
-            yield line
-
-
-anchor_cache = {}
-
-
-def anchors(path):
-    if path in anchor_cache:
-        return anchor_cache[path]
-    found = set()
-    counts = {}
-    for line in markdown_lines(path):
-        match = heading_pattern.match(line)
-        if not match:
-            continue
-        heading = re.sub(r"<[^>]+>", "", match.group(1)).replace("`", "")
-        base = re.sub(r"[^\w\- ]", "", heading.lower()).replace(" ", "-")
-        count = counts.get(base, 0)
-        counts[base] = count + 1
-        found.add(base if count == 0 else f"{base}-{count}")
-    anchor_cache[path] = found
-    return found
-
-
-failures = []
-for source in markdown_files:
-    text = "\n".join(markdown_lines(source))
-    for match in link_pattern.finditer(text):
-        target = match.group(1).strip()
-        if target.startswith("<") and target.endswith(">"):
-            target = target[1:-1]
-        if scheme_pattern.match(target) or target.startswith("//"):
-            continue
-        path_text, separator, fragment = target.partition("#")
-        destination = source if not path_text else source.parent / unquote(path_text)
-        try:
-            destination = destination.resolve()
-            destination.relative_to(root)
-        except ValueError:
-            failures.append(f"{source.relative_to(root)}: escapes repository: {target}")
-            continue
-        if not destination.exists():
-            failures.append(f"{source.relative_to(root)}: missing target: {target}")
-            continue
-        if separator and destination.suffix.lower() == ".md":
-            fragment = unquote(fragment).lower()
-            if fragment not in anchors(destination):
-                failures.append(f"{source.relative_to(root)}: missing anchor: {target}")
-
-if failures:
-    print("\n".join(failures), file=sys.stderr)
-    raise SystemExit(1)
-PY
-```
-
-### Repository Hygiene And Review
-
-Reject an active Go workspace, tracked workspace files, a vendor tree, or a
-module replacement:
-
-```sh
-case "$(go env GOWORK)" in
-    ''|off) ;;
-    *) printf '%s\n' 'an active Go workspace is not allowed' >&2; exit 1 ;;
-esac
-test -z "$(git ls-files go.work go.work.sum)"
-test ! -e vendor
-if grep -Eq '^[[:space:]]*replace([[:space:]]|\()' go.mod
-then
-    printf '%s\n' 'go.mod contains a replacement' >&2
-    exit 1
-fi
-```
-
-Check whitespace in both unstaged and staged changes. List ignored files and
-scan tracked content for common credential forms:
-
-```sh
+gofmt -w path/to/changed.go
 git diff --check
-git diff --cached --check
-test -z "$(git ls-files --others --ignored --exclude-standard)"
-credential_pattern='-----BEGIN ([A-Z0-9]+ )?PRIV''ATE KEY-----|AKI''A[0-9A-Z]{16}|gh[pousr]_[A-Za-z0-9]{36,}|sk-[A-Za-z0-9]{32,}'
-if git grep -nEI -e "$credential_pattern" -- .
-then
-    printf '%s\n' 'possible credential found' >&2
-    exit 1
-fi
 ```
 
+Release candidates use `./scripts/release.sh VERSION`, which runs the selected
+profile checks and publishes the tag. The [release procedure](release.md)
+describes its requirements and recovery. For an optional diagnostic of a
+committed candidate without publication, run `./scripts/releasekit.sh check
+VERSION`; this is not a required pre-publication step.
+
+The example workflows are covered by the Go suite; use
+`go test ./examples/go-library/...` to run them separately during development.
+The focused `python3 scripts/test-release.py` command exercises the release
+tooling against disposable local repositories; use it when changing that
+tooling, not as a recurring release gate. Tests and examples require no provider
+credentials or live inference calls.
+
+Ordinary validation can run on intended uncommitted work. Release validation
+uses committed candidate contents and the selected local Go installation.
+
+### Change Review
+
 Inspect `git status --short --untracked-files=all` and the complete diff before
-accepting a change. The status may contain only the intended source changes
-during development. Reject credentials, private keys, environment files,
-generated binaries, test or coverage output, downloaded assets, template
-residue, and any other artifact that does not belong in source control. The
-credential scan catches common forms but does not replace inspection of the
-actual change.
+accepting a change. Reject unintended files, credentials, private data,
+generated binaries, test output, downloaded assets, and template residue.
 
-After committing the accepted change, require a clean candidate:
+For releases, follow the [release procedure](release.md). Do not repeat its
+candidate checks as a separate sign-off. Roadmap retirement remains preparation
+housekeeping.
 
-```sh
-test -z "$(git status --porcelain)"
-```
+Follow the [agent commit policy](policy/agents.md#git-commit-policy) for local
+commits. Ordinary commits may leave unrelated work in the workspace; preserve
+and report it. A clean worktree and index are required for release candidates,
+as enforced by the publication command, not for every ordinary commit.

docs/formats.md

@@ -2,382 +2,442 @@
 
 ## Purpose
 
-This document is the canonical contract for Promptkit prompt-definition,
-profile, and schema files. The [Go package consumer guide](consumers/pkg-promptkit.md)
-explains how to select these sources and invoke the engine. The
+This is the canonical reference for Promptkit's backend, profile, prompt, and
+schema definitions. The [Go consumer guide](consumers/pkg-promptkit.md) shows
+how to select sources and use an engine; the
 [OpenAI-compatible integration contract](integrations/openai-compatible-chat.md)
-owns the resulting outbound wire behavior.
+owns outbound wire behavior. [ConfigKit](https://forge.maximumdirect.net/base-layer/configkit/src/tag/v1.1.0/docs/config.md)
+owns the generic YAML grammar, imports, layering, inheritance, and diagnostics.
+This reference describes what Promptkit does with the final effective values.
 
-Prompt and profile sources recursively discover files ending in `.yaml` or
-`.yml`. Each prompt-definition and profile file contains exactly one YAML
-document; comments and trailing whitespace are allowed. YAML decoding is
-strict: unknown fields are errors for the selected definition. Definitions are
-selected by their YAML `id`, not their file name or directory.
+A configuration document may contain top-level `backends`, `profiles`, and
+`prompts` lists, plus a `collections` mapping of namespaced resource groups.
+Each resource has an explicit literal `id`, not a filename-derived identity.
+IDs and references must be nonblank and remain exact: Promptkit does not trim or
+re-key them. Unknown fields are errors. Promptkit installs a default-on bundled
+OpenRouter underlay; callers still explicitly select a profile.
+
+```yaml
+backends:
+  - id: local
+    endpoint: http://localhost:8000/v1
+    concurrency_limit: 2
+profiles:
+  - id: local-summary
+    backend: local
+    model: example-model
+prompts:
+  - id: meeting.summary
+    default_profile: local-summary
+    messages:
+      - role: user
+        content: Summarize this note.
+    output:
+      format: text
+      validation_mode: none
+collections:
+  support:
+    release_version: "1.4.0"
+    backends:
+      - id: offline
+        endpoint: https://example.invalid/v1
+    profiles:
+      - id: base
+        backend: support/offline
+        model: offline-model
+      - id: fast
+        extends: {id: base}
+        timeout_seconds: 30
+    prompts:
+      - id: answer
+        default_profile: support/fast
+        messages:
+          - role: user
+            content: Answer the selected caller input.
+        output:
+          format: text
+          validation_mode: none
+```
+
+## Source Headers, Collections, And Resource IDs
+
+An input may omit a source header. When present, the source-root `version` is
+reserved for ConfigKit's specification header and accepts only the exact string
+`"2"` or strict decimal integer `2`; null and other values are invalid. This
+header is distinct from a collection's `release_version` and ConfigKit's own
+software version. ConfigKit checks and consumes accepted headers on non-main
+inputs; a declaration on the main input remains visible in explicit
+configuration inspection. Headerless inputs still undergo ordinary final-value
+validation. ConfigKit owns generic import and source-history behavior.
+For a standalone prompt item, a root `version` is only a source header, not
+prompt metadata.
+
+The root resource lists contain unqualified IDs. Each `collections` entry
+has a namespace key and may contain `release_version` plus the same three
+resource lists. Collection resource IDs are local keys; their runtime IDs
+are `<namespace>/<local-key>`. Root and collection keys may be the same. A
+namespace and local key must be nonblank and slash-free, and references must
+name their intended exact ID; mounting a source does not rewrite references.
+For example, the `support/fast` profile above stores `support/offline` as its
+backend ID. `extends: {id: base}` refers to a sibling in the same resource
+list. It is not a qualified or cross-collection reference.
+
+`release_version` is an optional opaque collection label. Absent or null means
+no label; an explicitly empty string is retained. It does not select a resource
+or change execution. Runtime lookups
+use the captured effective resource indexes. `InspectConfiguration` is an
+explicit operator view of resolved settings and provenance. Selected field
+normalizations and the derived queue default are recorded in that tree, with
+source history retained. Request-time completion remains separate. Source
+history is not a second runtime resolver, and the
+retained effective shape is not flattened or rewritten into those indexes.
+
+For examples of explicitly selecting native collection inputs and inspecting
+inventories, see the [consumer guide](consumers/pkg-promptkit.md) and its
+[maintained preparation example](../examples/go-library/prepare/main.go). For
+breaking identity and prompt-selection changes, see the
+[configuration migration guide](consumers/configuration-migration.md).
+
+Every effective backend, profile, and prompt definition is validated when
+`NewEngine` constructs the engine, even if it is never selected. A profile may
+be partial: an endpoint and model need only be complete after request overrides
+are applied. A prompt's default profile similarly need not be runnable without
+an overriding request. Missing request data, artifacts, schemas, credentials,
+and provider work remain runtime concerns.
+
+The engine captures effective definitions, referenced message-file contents,
+and static template syntax at construction. It does not reread configuration or
+message files on ordinary runs; construct a new engine to observe edits.
+Schema and request artifact reads still happen during preparation. Use
+`InspectConfiguration` for an explicit effective-configuration or provenance
+report, `InspectProfile` for backend-completed profile settings, and preparation
+for request-specific rendered details. For identity discovery before inspection,
+see the consumer guide's [resource and collection listing](consumers/pkg-promptkit.md#discover-available-resources-and-collections).
 
 ## Prompt Definitions
 
-A prompt definition describes inputs, Go-template messages, an optional
-default profile, and an output contract.
+The `prompts` list contains definitions selected by their literal IDs:
 
 ```yaml
-id: meeting.summary
-version: "1.0.0"
-default_profile: local-summary
-description: Summarize a synthetic meeting note.
-session_id: '{{.session}}'
-inputs:
-  - name: note
-    required: true
-    content_type: text/plain
-    description: Meeting note to summarize.
-messages:
-  - role: system
-    content: Return a concise summary.
-    cache_control:
-      type: ephemeral
-      ttl: 1h
-  - role: user
-    content_file: ./summary.user.md
-output:
-  format: markdown
-  validation_mode: basic
-  repair_attempts: 0
+prompts:
+  - id: meeting.summary
+    default_profile: local-summary
+    description: Summarize a synthetic meeting note.
+    session_id: '{{.session}}'
+    inputs:
+      - name: note
+        required: true
+        content_type: text/plain
+        description: Meeting note to summarize.
+    messages:
+      - role: system
+        content: Return a concise summary.
+        cache_control:
+          type: ephemeral
+          ttl: 1h
+      - role: user
+        content_file: ./summary.user.md
+    output:
+      format: markdown
+      validation_mode: basic
+      repair_attempts: 0
 ```
 
-| Field | Required | Meaning |
-| --- | --- | --- |
-| `id` | yes | Non-empty prompt identifier used by `RunRequest.PromptID`. |
-| `version` | yes | Non-empty version selected by an optional `RunRequest.PromptVersion`. |
-| `default_profile` | no | Non-empty profile ID used when the request omits `ProfileID`. |
-| `description` | no | Human-readable description. |
-| `session_id` | no | Go template rendered from request variables and input helpers. |
-| `inputs` | no | Declared input metadata. |
-| `messages` | yes | One or more chat-message templates. |
-| `output` | yes | Output format and validation settings. |
-
-When a request omits a version, the selected prompt ID must identify exactly
-one definition. When it supplies a version, the ID and version pair must be
-unique.
-
-Exact prompt inspection uses this same configured source, strict decoding,
-referenced content-file resolution, and ID/version selection. It reports the
-selected definition's declared metadata without changing the prompt format or
-executing the definition.
+`messages` must be nonempty and `output` must be a valid mapping. Optional
+`default_profile`, `description`, `session_id`, `inputs`, and cache-control
+fields may be absent or cleared with null where appropriate. A null whole
+definition, required role, output format, or validation mode is invalid.
+`required` and `repair_attempts` default to false and zero, respectively, and
+are not nullable. `InspectPrompt` selects the same captured definition without
+rendering or resolving a profile.
 
 ### Inputs
 
-Each `inputs` item has these fields:
-
-| Field | Required | Meaning |
-| --- | --- | --- |
-| `name` | yes | Non-empty name used by the request input map and `input` template helper. Names must be unique. |
-| `required` | no | When true, preparation fails if the request omits the input. The default is false. |
-| `content_type` | no | Expected media-type metadata. |
-| `description` | no | Human-readable input description. |
-
-Requests supply inputs as inline or file-backed `ArtifactRef` values. Declared
-required inputs must be present. A template reference also requires the named
-input to exist, whether or not it was declared. Extra request inputs are
-allowed.
+Each `inputs` item may contain `name`, `required`, `content_type`, and
+`description`. `name` is required, nonblank, and unique within the prompt.
+Omitted `required` means false. Null `inputs` means no declarations. Requests
+supply inline or file-backed `ArtifactRef` values; declared required inputs
+must be present. A template reference also needs its named input, even when
+undeclared. Extra request inputs are allowed.
 
 ### Messages And Templates
 
-Each message has a non-empty `role` and exactly one of:
+Each message needs a role and exactly one nonnull `content` or `content_file`.
+An explicitly empty inline `content` is valid. Null clears either content
+selection. Roles are trimmed and lowercased, then must be `developer`,
+`system`, `user`, or `assistant`. Static template syntax is checked during
+construction; missing variables and named inputs fail when rendered.
+Request variables are template data (`{{.audience}}`), and
+`{{input "note"}}` renders an input body.
 
-- `content`, containing an inline Go template; or
-- `content_file`, naming a file whose contents are the Go template.
+`content_file` uses the effective field's own source context, including when
+inherited or imported. Its relative path is resolved against that source's
+base directory within its rooted filesystem. Absolute paths and lexical
+escapes from the selected root fail. Host-file sources also prevent symlink
+escapes. Promptkit preserves the path text, including meaningful whitespace,
+after checking that it is not blank. A source without a unique usable resource
+context must supply one explicitly or override the field. Content is captured
+at construction, not reopened for each request.
 
-`content_file` must be a relative path. It resolves from the directory that
-contains the prompt file and must remain within the configured prompt source
-root; parent components are allowed only when the resolved target remains
-inside that root. Absolute paths and paths that escape the root are rejected.
-Operating-system directory and single-file sources also reject symlink targets
-outside the root, while injected `fs.FS` sources apply containment in that
-filesystem's relative path namespace. For `WithPromptFile`, the source root is
-the directory containing the selected prompt file. Promptkit uses the parsed
-path text exactly after checking separately that it is not blank, so leading
-and trailing whitespace can name real filesystem entries.
+Optional `session_id` uses the same template variables and input helper. Its
+rendered value is trimmed, omitted if empty, and limited to 256 Unicode code
+points. A nonblank direct request session ID bypasses the template.
 
-Request variables are the template data, so a variable named `audience` is
-referenced as `{{.audience}}`. The `{{input "note"}}` helper renders the body
-of a named input. Missing variables and input references are errors.
-
-The optional `session_id` uses the same template data and input helper. Its
-rendered value is trimmed, omitted when empty, and limited to 256 Unicode code
-points. A nonblank direct request session ID bypasses this template completely;
-a blank direct value leaves the template behavior unchanged.
+`InspectRequirements` can enumerate recognizable dependencies from captured
+templates before artifact content is available. Its conservative syntax
+coverage does not change template validity or rendering; see the [consumer
+guide](consumers/pkg-promptkit.md#inspect-requirements-before-preparation) and
+the method's [GoDoc](../requirements.go) for that inspection boundary.
 
 ### Cache Control
 
-`cache_control` is optional and has these fields:
-
-| Field | Required | Values |
-| --- | --- | --- |
-| `type` | yes | `ephemeral` |
-| `ttl` | no | Empty or `1h` |
-
-Promptkit preserves cache-control metadata on the rendered message. The
-outbound integration determines its wire representation.
+Optional `cache_control` has required `type: ephemeral` and optional `ttl`,
+which may be empty or `1h`. Promptkit retains this metadata on rendered
+messages; the integration contract defines its wire representation.
 
 ### Output Contract
 
 | Field | Required | Values or behavior |
 | --- | --- | --- |
-| `format` | yes | `text`, `markdown`, or `json`. |
-| `validation_mode` | yes | `none`, `basic`, `json`, or `json_schema`. |
-| `schema_path` | for `json_schema` | Path to a schema in the configured schema source. |
-| `repair_attempts` | no | Integer zero or greater; omitted means zero. |
+| `format` | yes | `text`, `markdown`, or `json` |
+| `validation_mode` | yes | `none`, `basic`, `json`, or `json_schema` |
+| `schema_path` | for `json_schema` | Path in the selected schema source |
+| `repair_attempts` | no | Integer 0–3; default 0. Positive values need `basic`, `json`, or `json_schema` |
 
-The validation modes behave as follows:
+`none` skips content validation; `basic` requires nonempty content; `json`
+accepts any valid JSON value; `json_schema` checks the selected schema.
+`format` determines output artifact metadata. JSON Schema mode also supplies
+structured-output metadata to compatible clients. Plain `json` does not
+request a provider-native JSON-object constraint. A positive repair budget
+counts additional generations after failed validation; exhaustion returns the
+last candidate and its failed validation result. Operational errors remain
+errors. A request-level `OutputContract` replaces the whole configured one,
+not individual fields; an empty request format defaults to `text`.
 
-- `none` skips content validation;
-- `basic` requires non-empty generated content;
-- `json` requires valid JSON; and
-- `json_schema` requires valid JSON that satisfies the selected schema.
+## Backend Definitions
 
-`format` controls output artifact metadata. JSON Schema mode also supplies the
-schema to compatible model clients as structured-output metadata. The public
-engine does not install an output repairer, so its validation is single-pass
-even when a positive `repair_attempts` value is present.
+Backend definitions are supplied through caller-selected configuration
+sources. The conventional backend IDs are names only; PromptKit does not
+install definitions for them.
 
-A request-level `OutputContract` replaces the complete prompt output contract.
-It does not merge individual fields. If its format is empty, Promptkit uses
-`text`.
+A backend definition may have `endpoint`, `api_key_env`,
+`concurrency_limit`, `queue_capacity`, `extra_params`, and `smoke_tests`. The
+final endpoint is required and must be an absolute HTTP(S) URL with a host, no
+user information, query, or fragment. An omitted or null API-key environment
+name means no lookup. Omitted/null concurrency is zero (unlimited); a supplied
+negative value is invalid. For positive concurrency, omitted/null queue
+capacity defaults to 1024. An explicit queue capacity, including zero,
+requires positive concurrency and must be nonnegative. Unlimited concurrency
+has no queue. Provider maps contain JSON-compatible values; null entries are
+ordinary JSON null. Unknown fields, including a raw `api_key`, fail.
 
-## Built-In Backends
+`smoke_tests` is optional backend metadata for explicit diagnostic operations.
+It may contain `default_profile`, `reachability`, and `authentication`:
 
-Every engine provides these reserved OpenAI-compatible backend IDs. Consumers
-must not register either ID with `WithBackend`; exact registration and
-reservation behavior belongs to the [`Backend` GoDoc](../backends.go).
+```yaml
+backends:
+  - id: openrouter
+    endpoint: https://openrouter.ai/api/v1
+    smoke_tests:
+      default_profile: application-profile
+      reachability:
+        endpoint: https://status.example/ready?check=1
+      authentication:
+        kind: openrouter_key
+        endpoint: https://openrouter.ai/api/v1/key
+```
 
-| ID | Base endpoint | API-key environment variable | Active generation limit | Default queue capacity |
-| --- | --- | --- | ---: | ---: |
-| `openrouter` | `https://openrouter.ai/api/v1` | `OPENROUTER_API_KEY` | 16 | 1024 |
-| `rakestrawhome` | `https://inference.ai.rakestrawhome.com/v1` | `RAKESTRAWHOME_INFERENCE_API_KEY` | 4 | 1024 |
+An omitted or null `smoke_tests` block disables these metadata defaults.
+`default_profile` is preserved as an exact, nonblank profile ID; its existence
+is checked only if an inference diagnostic selects that default. Omitted/null
+`default_profile` means there is no inference default. Omitted/null
+`reachability` or its endpoint uses the effective backend endpoint. An explicit
+reachability endpoint is trimmed and must be an absolute HTTP(S) URL with a
+host; unlike the backend base endpoint, its path and query are allowed. Null
+means fallback, while an explicit blank value is invalid.
+
+Omitted/null `authentication` means no authentication adapter is declared. A
+non-null object requires `kind`, which is trimmed and case-normalized; the
+recognized values are `none` and `openrouter_key`. Its omitted/null endpoint
+uses the adapter's backend-relative destination (`openrouter_key` appends
+`/key` to the effective backend base); an explicit endpoint follows the same
+full-URL rules as reachability. `none` declares authentication inapplicable.
+These values are configuration only: engine construction does not contact a
+backend.
+
+These fields are resolved through the ordinary ConfigKit source layers before
+Promptkit validates the final effective values. Sibling inheritance can overlay
+individual nested fields; a later same-ID resource replaces the whole record.
+An explicit null can clear the whole `smoke_tests` block or one of its nested
+objects. An endpoint set to null uses
+its documented fallback, while an explicit blank endpoint is invalid. The
+metadata does not change ordinary generation credential requirements or
+provider settings.
 
 ## Profile Definitions
 
-A profile supplies model execution settings:
+Profiles have literal IDs and may inherit a sibling parent with
+`extends`:
 
 ```yaml
-id: local-summary
-backend: openrouter
-model: example-model
-temperature: 0.2
-max_tokens: 500
-top_p: 0.95
-timeout_seconds: 90
-service_tier: flex
-reasoning_effort: medium
-extra_params:
-  provider_option: enabled
+profiles:
+  - id: local-summary
+    backend: local
+    model: example-model
+    temperature: 0.2
+    max_tokens: 500
+    timeout_seconds: 90
+    extra_params:
+      provider_option: enabled
+  - id: local-summary-fast
+    extends: {id: local-summary}
+    timeout_seconds: 30
+    reasoning_effort: low
 ```
 
-A derived profile can use a named base and override only the settings it owns:
-
-```yaml
-id: local-summary-fast
-base_profile: local-summary
-timeout_seconds: 30
-reasoning_effort: low
-```
-
-| Field | Required | Meaning |
-| --- | --- | --- |
-| `id` | yes | Profile identifier, trimmed before selection and publication. It must be non-empty after trimming and unique within one source after normalization. |
-| `base_profile` | no | One optional parent profile ID. A derived profile may inherit target fields from it. |
-| `backend` | unless `endpoint` is present | Backend registry ID. It is trimmed and registry membership is checked when the profile is prepared or inspected. |
-| `endpoint` | unless `backend` is present | OpenAI-compatible base URL, including an API version path when required. A nonempty value is trimmed and must be absolute HTTP or HTTPS with a host and without user information, a query, or a fragment. When both connection fields are present, this overrides the backend endpoint without changing backend identity. |
-| `model` | yes | Non-empty provider model name. |
-| `temperature` | no | Number from 0 through 2. |
-| `max_tokens` | no | Integer zero or greater. |
-| `top_p` | no | Number from 0 through 1. |
-| `timeout_seconds` | no | Per-generation deadline in whole seconds; integer zero or greater. |
-| `service_tier` | no | Provider-specific request tier. |
-| `reasoning_effort` | no | Provider-specific reasoning setting. |
-| `api_key_env` | no | Optional environment-variable lookup source for an API key. |
-| `extra_params` | no | JSON-compatible provider-specific outbound fields. |
-
-Raw `api_key` is prohibited in profile YAML. Store only an environment
-variable name in `api_key_env`.
-
-A standalone profile must provide a model and at least one of `backend` or
-`endpoint`. A derived profile may omit those target fields because its selected
-base chain can provide them. Local parsing still validates a derived profile's
-own ID, supplied endpoint, execution-setting bounds, and `extra_params`.
-
-Promptkit does not infer a backend from a model or endpoint. Endpoint-only
-profiles remain supported and have no effective backend ID.
-The engine always provides the built-in `openrouter` and `rakestrawhome` IDs.
-Consumers can add engine-scoped IDs with
-[`WithBackend`](../backends.go); exact registration validation belongs to its
-GoDoc.
-
-`extra_params` accepts null, booleans, finite numbers, strings, arrays, and
-objects with string keys. Keys must be non-empty. With the built-in client,
-they also cannot collide with the standard fields listed in the
-[outbound request contract](integrations/openai-compatible-chat.md#request-body).
-Excessively deep or large JSON-shaped values are rejected for safety.
+Profiles may also supply `endpoint`, `top_p`, `service_tier`, `api_key_env`,
+and `api_key_required`. Promptkit does not infer a backend from an endpoint or
+model. A profile with only an endpoint has no backend capacity identity.
+Nonblank endpoints follow the backend URL rules; temperature is 0–2, top_p
+is 0–1, max_tokens and timeout_seconds are nonnegative. Provider maps accept
+JSON-compatible values with nonempty keys and cannot collide with the built-in
+client's [standard request fields](integrations/openai-compatible-chat.md#request-body).
+Credential values do not belong in profiles: raw `api_key` is invalid.
 
 ### Defaults And Overrides
 
-Execution settings resolve in this order:
+After effective profile inheritance, Promptkit applies the selected backend's
+defaults once, then request `ExecutionTargetOverride` values. Backend-derived
+settings are not inheritance input: changing a child profile's backend does
+not carry the parent's derived endpoint or credential source into it. Changing
+only a model or backend does not clear unrelated explicitly inherited fields.
 
-1. the framework timeout baseline;
-2. the selected backend, when the profile names one;
-3. the selected profile; and
-4. request `ExecutionTargetOverride` values.
+| Profile field | Absent | Explicit value | Explicit null |
+| --- | --- | --- | --- |
+| `backend` | No backend | Exact backend ID; blank means none | No backend |
+| `endpoint` | Selected backend endpoint | Valid nonblank URL; empty clears | Clear, without backend fallback |
+| `model` | Unspecified | Nonblank model; empty clears | Unspecified |
+| `temperature`, `top_p`, `max_tokens` | Unspecified | Valid value, including outbound zero | Suppress outbound control |
+| `service_tier`, `reasoning_effort` | Unspecified | Normalized value; blank suppresses | Suppress outbound control |
+| `timeout_seconds` | 600 seconds | Nonnegative; zero disables generation deadline | Reset to 600 seconds |
+| `api_key_env` | Backend lookup name, if any | Trimmed name; blank disables | Disable backend lookup |
+| `api_key_required` | False | Boolean, including false | False |
+| `extra_params` | Backend defaults | Recursively overlay profile map on backend map | Suppress configured provider map |
 
-The framework baseline is:
+An empty mapping does not clear inherited map entries; null clears the whole
+map. Nested maps merge recursively, while scalar and array values replace.
+Runtime nonempty `ExtraParams` overrides replace the entire configured map.
+Request overrides otherwise follow the [public value GoDoc](../types.go).
+For example, a runtime `extra_params["extends"]` key is ordinary provider
+data, not a configuration inheritance instruction. An explicit zero timeout
+does not disable the caller context or HTTP transport timeout.
 
-| Setting | Default |
+### Bundled OpenRouter Configuration
+
+The default bundle installs root backend `openrouter`, endpoint
+`https://openrouter.ai/api/v1`, optional credential environment name
+`OPENROUTER_API_KEY`, and smoke metadata selecting `gpt-luna-latest` with the
+`openrouter_key` authentication adapter. Existing diagnostic endpoint fallback
+applies. No prompt, implicit profile selection, or tuning defaults are installed.
+Disable the entire bundle with `WithBundledOpenRouter(false)`.
+
+The root profile IDs and their models are:
+
+| Profile | Model |
 | --- | --- |
-| `temperature` | Unspecified and omitted from compatible provider requests unless a profile or runtime override selects it. |
-| `max_tokens` | Unspecified and omitted from compatible provider requests unless a profile or runtime override selects it. |
-| `top_p` | Unspecified and omitted from compatible provider requests unless a profile or runtime override selects it. |
-| `timeout_seconds` | `600` |
+| `gpt-astra-latest` | `~openai/gpt-astra-latest` |
+| `gpt-sol-latest` | `~openai/gpt-sol-latest` |
+| `gpt-terra-latest` | `~openai/gpt-terra-latest` |
+| `gpt-luna-latest` | `~openai/gpt-luna-latest` |
+| `claude-fable-latest` | `~anthropic/claude-fable-latest` |
+| `claude-opus-latest` | `~anthropic/claude-opus-latest` |
+| `claude-sonnet-latest` | `~anthropic/claude-sonnet-latest` |
+| `claude-haiku-latest` | `~anthropic/claude-haiku-latest` |
+| `gemini-pro-latest` | `~google/gemini-pro-latest` |
+| `gemini-flash-latest` | `~google/gemini-flash-latest` |
+| `auto` | `openrouter/auto` |
+| `free` | `openrouter/free` |
 
-Numeric zero in a file or in-memory profile does not select a numeric value.
-For `temperature`, `max_tokens`, and `top_p`, it leaves the provider control
-unspecified. For `timeout_seconds`, it retains the framework deadline. Numeric
-request overrides use pointers, so an explicit zero is retained and sent to
-compatible providers. In particular, an explicit request `timeout_seconds` of
-zero disables the per-generation deadline while leaving the caller context and
-transport timeout intact.
+The profiles `auto-low`, `auto-medium`, `auto-high`, `auto-xhigh`, and
+`auto-max` extend `auto` and add an `auto-router` plugin with the corresponding
+`cost_tier`. Replacing `auto` changes their inherited backend and model while
+retaining their plugins. These are root IDs even when a model contains `/`.
+For example, a caller can replace `auto` as a complete record and extend another
+bundled profile under a new ID:
 
-Non-empty profile strings replace backend defaults, and non-empty request
-strings replace both. Request reasoning is the exception: a nil
-`ReasoningEffort` pointer inherits the profile, a pointer to a nonblank string
-trims and replaces it, and a pointer to a blank string clears it. Backend
-identity is retained when either layer overrides the endpoint, so the override
-also retains any engine-local capacity policy configured for that backend.
-Capacity configuration belongs to the Go
-[`Backend` API](../backends.go), not prompt or profile YAML. A non-empty
-`extra_params` map at each layer replaces the entire lower-precedence map
-rather than merging keys.
-The [outbound integration contract](integrations/openai-compatible-chat.md)
-defines how the effective settings are serialized.
+```yaml
+profiles:
+  - id: auto
+    backend: openrouter
+    model: openrouter/free
+  - id: my-luna
+    extends: {id: gpt-luna-latest}
+    timeout_seconds: 60
+```
+
+See OpenRouter's [alias examples](https://openrouter.ai/~openai/gpt-luna-latest),
+[auto router and cost tiers](https://openrouter.ai/docs/guides/routing/routers/auto-router),
+and [free router](https://openrouter.ai/docs/guides/routing/routers/free-router)
+for provider behavior and availability.
 
 ### Source And Profile Precedence
 
-An explicit request profile ID takes precedence over the prompt's
-`default_profile`. If neither is present, preparation fails. Exact profile
-inspection instead takes one explicit profile ID and does not use a prompt
-default.
+`NewEngine` composes the bundled OpenRouter underlay first by default, then
+caller underlays in their given order, followed by an
+optional main input. Later layers overlay earlier ones. Inputs within one
+underlay group are strictly composed: conflicting contributions are errors,
+not priority signals. The directory helpers discover `.yaml` and `.yml` files recursively
+in sorted order, but that order does not give one file higher precedence.
+An explicitly selected missing/unreadable root fails; an empty root contributes
+nothing. Imports are disabled by these helpers unless the caller enables them
+on an input. Single files can use ConfigKit inputs directly.
 
-Profile sources resolve matching IDs in this order:
+Root backends, profiles, and prompts, as well as collection declarations and
+their resource lists, follow these same source rules. `CollectionInput`
+mounts one explicitly selected source at a collection destination and enables
+ConfigKit imports for that input; it does not discover or select sources.
+A later same-ID record replaces the lower definition in full. Use a distinct
+inherited sibling when a variant should retain its parent's fields.
+A request `ProfileID` takes precedence over a prompt `default_profile`; if
+neither is selected, preparation fails. Exact profile inspection selects one
+explicit ID and does not consult a prompt default.
 
-1. in-memory profiles supplied with `WithProfiles`;
-2. the ordinary configured source selected by a profile file, `fs.FS`, or
-   configured profile directory;
-3. application fallback profiles supplied with `WithFallbackProfileFS`; and
-4. embedded built-in profiles.
+### Resource Inheritance
 
-A profile source supplies a complete definition; definitions and their fields
-are not merged across sources. A higher-precedence source falls back only when
-the requested profile ID is absent. An invalid matching profile is an error and
-does not fall back. In-memory `Profile` values follow the same ranges as YAML
-profiles. They use `APIKeyRequired` for request-scoped credentials instead of
-`api_key_env`. Preparation and exact profile inspection use this same source
-precedence.
-
-When a selected definition names `base_profile`, every profile ID in that
-chain is looked up through this same precedence order. A higher-precedence
-definition therefore shadows a lower-precedence definition of the same base
-ID, including a built-in. References are not source-qualified.
-
-### Profile Inheritance
-
-Promptkit resolves one linear base chain of at most 32 profiles, including the
-selected profile. It merges settings from the root base to the selected leaf.
-The leaf's `id` remains the selected profile identity. Nonblank string fields
-(`backend`, `endpoint`, `model`, `service_tier`, `reasoning_effort`, and
-`api_key_env`) and nonzero numeric fields replace inherited values. A nonempty
-`extra_params` map replaces the complete inherited map rather than merging
-keys, and `APIKeyRequired: true` remains true through the chain. Backend and
-endpoint are independent: replacing one does not clear the other.
-
-There is no profile-level clearing syntax. Blank strings, zero numbers, false,
-and empty maps remain unspecified and inherit from a base. Use existing
-presence-aware request overrides where an execution needs an explicit zero or
-empty reasoning setting.
-
-An absent directly selected profile reports the ordinary not-found error. Once
-the selected profile exists, a missing base, cycle, overlong chain, or
-incomplete resolved target is a profile-load failure. Ordinary operations
-resolve chains afresh; prepared execution retains the fully resolved target.
-
-## Built-In Profile Catalog
-
-Every built-in profile selects one maintained built-in backend and inherits
-that backend's connection and credential metadata. Profile files do not repeat
-those values. A configured, application fallback, or in-memory profile with
-the same profile ID takes precedence.
-
-| Provider | ID | Backend | Model |
-| --- | --- | --- | --- |
-| aion-labs | `aion-2` | `openrouter` | `aion-labs/aion-2.0` |
-| anthropic | `claude-fable-latest` | `openrouter` | `~anthropic/claude-fable-latest` |
-| anthropic | `claude-haiku-latest` | `openrouter` | `~anthropic/claude-haiku-latest` |
-| anthropic | `claude-opus-latest` | `openrouter` | `~anthropic/claude-opus-latest` |
-| anthropic | `claude-sonnet-latest` | `openrouter` | `~anthropic/claude-sonnet-latest` |
-| deepseek | `deepseek-3-2` | `openrouter` | `deepseek/deepseek-v3.2` |
-| deepseek | `deepseek-4-flash` | `openrouter` | `deepseek/deepseek-v4-flash` |
-| deepseek | `deepseek-4-pro` | `openrouter` | `deepseek/deepseek-v4-pro` |
-| google | `gemini-2-flash` | `openrouter` | `google/gemini-2.5-flash` |
-| google | `gemini-2-flash-lite` | `openrouter` | `google/gemini-2.5-flash-lite` |
-| google | `gemini-2-pro` | `openrouter` | `google/gemini-2.5-pro` |
-| google | `gemini-3-flash-lite` | `openrouter` | `google/gemini-3.1-flash-lite` |
-| google | `gemini-flash-latest` | `openrouter` | `~google/gemini-flash-latest` |
-| google | `gemini-pro-latest` | `openrouter` | `~google/gemini-pro-latest` |
-| google | `gemma-4-31b` | `openrouter` | `google/gemma-4-31b-it:exacto` |
-| google | `rakestrawhome-gemma-4-31b` | `rakestrawhome` | `google/gemma-4-31b-it` |
-| minimax | `minimax-m2` | `openrouter` | `minimax/minimax-m2.5` |
-| minimax | `minimax-m3` | `openrouter` | `minimax/minimax-m3` |
-| mistral | `mistral-large-2512` | `openrouter` | `mistralai/mistral-large-2512` |
-| mistral | `mistral-medium-3-5` | `openrouter` | `mistralai/mistral-medium-3-5` |
-| mistral | `mistral-small-3` | `openrouter` | `mistralai/mistral-small-3.2-24b-instruct` |
-| mistral | `mistral-small-4` | `openrouter` | `mistralai/mistral-small-2603` |
-| nvidia | `nemotron-3-ultra` | `openrouter` | `nvidia/nemotron-3-ultra-550b-a55b` |
-| openai | `gpt-5-mini` | `openrouter` | `openai/gpt-5.4-mini` |
-| openai | `gpt-5-nano` | `openrouter` | `openai/gpt-5.4-nano` |
+ConfigKit resolves `extends: {id: parent}` for backends, profiles, and prompts.
+Each child inherits a surviving sibling in the same root or collection list;
+collection inheritance cannot cross namespaces. Parents are ordinary validated
+resources, and profiles may remain partial until request completion. Presence, explicit
+null, and recursively merged maps determine the final fields. Other resource
+references are exact IDs and use their qualified form when they select a
+collection resource. There is no Promptkit-specific depth limit or second
+base-chain resolver. Missing references and cycles fail at construction. For
+generic inheritance and layer mechanics, consult
+[ConfigKit](https://forge.maximumdirect.net/base-layer/configkit/src/tag/v1.1.0/docs/config.md).
 
 ## Schemas
 
-Schemas are JSON documents selected by a prompt or request
-`schema_path`. For a directory or `fs.FS` source, paths resolve within the
-configured source root. Referenced nested schemas resolve relative to the
-owning schema document. `WithSchemaFile` exposes one schema, addressed by its
-base name.
-
-An unreadable, invalid, or unresolvable schema produces an operational
-validation error. Generated content that is valid JSON but does not satisfy the
-schema produces a failed validation result.
+Schemas are JSON documents selected by a prompt or request `schema_path`.
+Directory and `fs.FS` schema paths resolve within the configured root;
+referenced nested schemas resolve relative to the owning schema document.
+`WithSchemaFile` exposes one schema under its base name. An unreadable,
+invalid, or unresolvable schema causes an operational validation error.
+Valid JSON that fails the schema instead yields a failed validation result.
 
 ## Credentials
 
-Credential values belong at the request or environment boundary, never in
-prompt, profile, schema, or example files:
+Only environment-variable names belong in backend/profile definitions, never
+credential values. A request may supply a direct `APIKey` or an `APIKeyEnv`
+override. A direct key takes precedence, followed by a request environment
+source, an explicit profile source, and an inherited backend source. A true
+profile `api_key_required` suppresses the backend-derived environment source,
+but an explicit profile or request environment source remains usable. False
+restores ordinary optional-credential behavior. Missing or blank optional
+environment values leave the built-in client's Authorization header absent;
+required availability is checked at preparation and again during prepared
+execution. Injected clients own their credential-resolution behavior.
 
-- a backend or file profile can name an optional environment lookup source
-  with `APIKeyEnv` or `api_key_env`;
-- an in-memory profile may set `APIKeyRequired` as an explicit local
-  requirement;
-- a request can provide a direct `APIKey` or override the optional `APIKeyEnv`
-  source; and
-- a direct request key takes precedence over environment lookup.
-
-After a direct request key, the credential-source precedence is request
-`APIKeyEnv`, profile `api_key_env`, then the backend default. An in-memory
-profile with `APIKeyRequired` clears an inherited backend environment name and
-requires a direct key unless the request explicitly supplies `APIKeyEnv`.
-Named environment sources are optional: when the selected source is absent,
-empty, or whitespace-only, the built-in client omits the `Authorization`
-header and handles the provider response normally. `APIKeyRequired` is the
-only explicit local availability requirement. Promptkit validates required
-credential availability during preparation and rechecks it when a prepared
-execution runs. Injected clients receive resolved source metadata but define
-their own credential-resolution behavior.
-Direct keys are excluded from JSON results and redacted by public string
-formatters. Environment-variable names may appear in prepared metadata, but
-their values do not.
+Credential values are excluded from JSON results and redacted by public
+string formatters. ConfigKit inspection redacts credential-shaped dynamic
+provider keys, but does not make arbitrary prompt/provider content safe to
+disclose. Applications own report disclosure and secret handling.

docs/integrations/openai-compatible-chat.md

@@ -45,6 +45,45 @@ precedence over the backend default. Only the name reaches prepared metadata;
 the environment value is read just before the provider call and is never added
 to the JSON body.
 
+## Explicit Smoke Diagnostics
+
+Promptkit's explicit smoke diagnostics are separate consumer operations; they
+are not part of ordinary `Run`, the offline examples, or release validation.
+With the built-in client, inference diagnostics use the same configured client
+and `/chat/completions` request as generation. Their content cases are private
+to Promptkit and are not application prompt, schema, or provider protocol
+extensions. The public
+[`Engine.SmokeTest` GoDoc](../../smoke_diagnostics.go) owns request selection,
+report facts, and injected-client behavior.
+
+HTTP-only diagnostics use the configured built-in HTTP client. Reachability is
+one unauthenticated `GET` to the selected diagnostic URL; an observed response
+counts as reachable even when its status is non-2xx. Authentication behavior
+is selected from backend metadata, not inferred from the host. The
+`openrouter_key` adapter makes one `GET` to `/key` relative to the effective
+backend base, unless configuration supplies a full diagnostic endpoint (see
+the [backend format reference](../formats.md#backend-definitions)). It may send
+the available bearer credential, but can dispatch without one. A 401 is
+recognized as rejection. A 200 is recognized as acceptance only when its JSON
+object has a `data` object containing string `label`, Boolean `is_free_tier`,
+and numeric `usage`; unrelated fields and account limits are ignored. This is
+only a minimal identifying subset of OpenRouter's documented
+[current-key response](https://github.com/OpenRouterTeam/docs/blob/main/api_reference/limits.mdx#checking-your-limits),
+not validation of the full account document. Other statuses and unrecognized
+200 responses do not establish acceptance.
+
+Diagnostic HTTP evidence describes the last request observed at Promptkit's
+RoundTripper boundary and any response associated with it. It includes the
+selected destination without URL user information, response status, and
+credential presence when observed. Redirects can change the last observed
+destination or strip credentials. This evidence does not prove that bytes
+reached a peer or reveal dispatch hidden inside a custom transport. Diagnostic
+reports do not retain headers, credentials, provider
+bodies, or generated text. A consumer-injected `LLMClient` can support inference
+diagnostics, but does not establish HTTP status, destination, reachability, or
+upstream credential-supply facts; Promptkit does not fall back to a built-in
+transport for HTTP-only diagnostics.
+
 ## Request Body
 
 The request body always contains `model` and `messages`. The execution
@@ -55,6 +94,12 @@ Each ordinary message contains its `role` and string `content`. A
 cache-controlled message instead uses a text content block containing `type`,
 `text`, and `cache_control`; an empty cache-control TTL is omitted.
 
+Promptkit sends only `developer`, `system`, `user`, and `assistant` roles and
+does so without provider-specific translation. Tool and deprecated function
+payloads are outside this text-message contract. A backend or model that
+rejects an otherwise supported role or context returns its ordinary provider
+error, which follows the normal generation-error path.
+
 The effective direct or prompt-rendered session ID is trimmed, limited to 256
 Unicode code points, and sent when nonempty as top-level `session_id`. It is
 never also sent as a session header.
@@ -67,13 +112,13 @@ The client conditionally includes:
 - non-empty `service_tier` and effective `reasoning_effort`; an explicitly
   disabled reasoning setting is empty and therefore omitted; and
 - `response_format` for JSON Schema structured output, including its name,
-  strict flag, and schema document.
+  strict flag, and schema document. Plain JSON validation does not add an
+  object-only response constraint.
 
 The engine resolves backend, profile, and request extra-parameter maps by
 whole-map replacement rather than key merging. The resulting effective map is
-then merged directly into the top-level body after JSON serialization is
-verified. Empty keys and collisions with these reserved fields are rejected
-before any provider call:
+merged into the top-level body. Empty keys, values that cannot be encoded, and
+collisions with these reserved fields are rejected before any provider call:
 
 - `model`
 - `session_id`
@@ -99,10 +144,20 @@ drained.
 
 The bounded body must contain exactly one OpenAI-compatible JSON response
 object followed only by JSON whitespace and EOF. The client returns the first
-choice's non-empty message content and maps prompt, completion, total, cached,
-and cache-write token counts. Invalid or truncated JSON, trailing non-whitespace
-data, a second JSON value, absent choices, empty first-choice content, and size
-overflow are malformed responses and return no partial result.
+choice's explicitly present string message content, including an empty or
+whitespace-only string, and maps prompt, completion, total, cached, and
+cache-write token counts. Invalid or truncated JSON, trailing non-whitespace
+data, a second JSON value, absent choices, missing content, `null` content,
+non-string content, and size overflow are malformed responses and return no
+partial result.
+
+Each of the five token fields retains whether the provider reported a
+non-null integer, including zero. Omitted and `null` fields are unknown for
+observations. A strictly decoded successful envelope can yield observed usage
+even when a later missing choice or content makes the attempt malformed.
+Invalid framing and non-2xx error bodies supply no observed usage; provider
+work may still have occurred. Reported counts are not inferred, reconciled,
+or proof of complete cost accounting.
 
 For a non-2xx status, Promptkit recognizes one JSON document with a top-level
 object-valued `error` member. Its optional `message` and `type` fields must be
@@ -130,6 +185,33 @@ identity and the exact transport error for `errors.Is` and `errors.As` checks.
 The rendered error does not include the selected endpoint, request headers,
 request content, credentials, or provider body.
 
+## Transient Status Retries
+
+The built-in path retries only HTTP 429, 500, 502, 503, and 504. It makes two
+additional attempts by default; engine configuration or an individual run can
+choose a different nonnegative count, including zero for single-attempt
+behavior. The same budget applies separately to initial generation and each
+structural repair generation. An injected `LLMClient` is dispatched once by
+Promptkit regardless of this setting. A custom `http.Client` on the built-in
+path remains subject to these retries.
+
+Each retry uses equal jitter between half and all of an exponential local
+delay, starting at 500 milliseconds and capped at 8 seconds. The first
+`Retry-After` header value sets a minimum not-before time measured from
+response receipt if it is valid unsigned decimal seconds or an HTTP date.
+Invalid values use local delay; past dates impose no extra delay. The provider
+minimum is not capped at 8 seconds. Retry waiting is cancellable and holds no
+active backend permit. A caller without a deadline can wait a long time for
+provider timing.
+
+The request body, selected URL, and resolved authorization are fixed across
+attempts of one generation. Each attempt creates a fresh HTTP request and
+receives its own configured generation and HTTP-client timeout after capacity
+admission. The caller context covers every attempt and wait without a reset.
+Transport failures, malformed responses, noneligible statuses, and invalid
+generated content are not transient retries. A positive budget ending in
+eligible statuses reports retry exhaustion with the final provider cause.
+
 ## Timeout And Cancellation
 
 Timeouts are layered:

docs/internal/capacity.md

@@ -14,7 +14,7 @@ stable JSON values.
 
 ## Construction And Pool Lifecycle
 
-Each root `NewEngine` call obtains a normalized capacity-policy snapshot from
+Each internal engine construction obtains a normalized capacity-policy snapshot from
 its immutable backend registry and constructs a new `Manager`. The manager
 creates one pool for each limited backend ID. It has no package-global mutable
 state, background workers, shutdown protocol, or persistence, so engines with
@@ -54,38 +54,59 @@ reserve another bounded slot.
 
 ## FIFO Generation Permits
 
-`NewClient` wraps the engine's selected internal model client after public
-client adaptation or built-in client construction. Initial generation and the
-default repairer receive the same wrapper. Their requests retain the same
-effective backend, credential, numeric-presence metadata, and structured-output
-settings, so scheduling does not change provider omission semantics between
-calls.
+The engine supplies one [generation coordinator](generation.md) to initial
+generation and the default repairer. Their requests retain the same effective
+backend, credential, numeric-presence metadata, structured-output settings,
+and retry budget. Scheduling does not change provider omission semantics
+between calls.
 
-For each `Generate` call, the wrapper selects a pool from the request's
-effective backend ID. An unlimited call passes directly to the next client. A
-limited call acquires an active permit, invokes the next client, and defers
-permit release so ordinary returns and panic unwinding both restore capacity.
+For each model attempt, the coordinator selects a pool from the request's
+effective backend ID. An unlimited call uses the unrestricted path. A limited
+call acquires an active permit and defers release around that attempt, so
+ordinary returns and panic unwinding both restore capacity.
 Preparation and validation never hold an active permit.
 
+The manager's `Acquire` operation owns that active-permit acquisition and
+returns an idempotent release function. The coordinator uses it around
+individual attempts. The earlier one-shot wrapper still delegates to the same
+operation for focused internal callers and tests, but engine assembly no
+longer uses it. There is one queue and one active limit per limited backend pool.
+
 When all active permits are occupied, calls join a mutex-protected FIFO waiter
 list. Releasing a permit transfers it directly to the oldest remaining waiter
 before making it generally available. Pools do not order work relative to
 other backend IDs.
 
-The wrapper passes generation requests, responses, and collaborator errors
-through unchanged. It owns scheduling only; the concrete model client remains
-responsible for provider transport behavior. The runner, rather than the
+The coordinator does not retry injected clients; it normalizes a nil successful
+response as malformed generation. The built-in HTTP client remains responsible
+for provider transport behavior. The runner, rather than the
 capacity layer, sums all five usage fields from the initial response and every
 completed repair response into the successful run result.
 
+An invocation-local observer receives attempt events after active permit
+release, including canceled acquisition without dispatch. A retry wait holds
+the logical admission lease but no active permit. Observer callbacks must not
+block on work that requires the same run's admission slot.
+
+## Smoke-Test Capacity
+
+The engine's smoke tester shares this manager and generation coordinator with
+the ordinary runner. Inference diagnostics reserve the same logical admission
+slot and acquire the same backend-scoped active permits for actual model
+attempts, including retries. They therefore participate in the same configured
+backend limits as `Run`. Reachability and authentication HTTP probes do not
+enter generation admission or consume active permits. The public
+[`Engine.SmokeTest` GoDoc](../../smoke_diagnostics.go) owns the caller-visible
+operation and injected-client limits.
+
 ## Cancellation And Release
 
 Admission checks the caller context before reserving a slot. A call canceled
 while waiting for an active permit removes its waiter under the same pool lock
-used to grant permits. If cancellation removes the waiter first, the wrapped
-client is not invoked. If a concurrent grant wins first, the call owns the
-permit and invokes the client with the original context, allowing the client
-to observe cancellation normally.
+used to grant permits. If cancellation removes the waiter first, no model
+attempt is dispatched. If a concurrent grant wins first, the call owns the
+permit; the coordinator checks the context before dispatch and releases the
+permit if cancellation has already occurred.
 
 This grant-or-cancel decision prevents lost and double-released permits.
 Admission leases and active permits are released after success, collaborator
@@ -111,8 +132,13 @@ release. The
 [external package capacity tests](../../capacity_contract_test.go) own the
 assembled public-engine behavior for configured limits, capacity errors,
 endpoint identity, engine independence, and injected clients. The
+[reliable execution contract tests](../../reliable_execution_contract_test.go)
+cover shared and independent backend pools during retry and repair. The
 [prepared-execution contract tests](../../prepared_execution_contract_test.go)
 own the public prepared-capacity boundary. The
 [root error-boundary tests](../../errors_internal_test.go) own preservation of
 the public generation category and context identity when generation is
-canceled.
+canceled. The assembled
+[smoke diagnostic contract tests](../../smoke_integration_contract_test.go)
+verify that inference diagnostics share the engine's Run admission and recover
+capacity after cancellation or failure.

docs/internal/generation.md

@@ -0,0 +1,63 @@
+# Internal Generation Coordination
+
+`internal/generation` coordinates one model-generation request through the
+existing engine-local [capacity manager](capacity.md) and the
+[built-in HTTP client](llm.md). The engine supplies the same coordinator to
+initial generation and the default output repairer.
+
+The coordinator has explicit built-in and injected-client constructors. The
+built-in path prepares the provider request once, then acquires and releases an
+active backend permit around each actual HTTP attempt. Only HTTP 429, 500, 502,
+503, and 504 can lead to another attempt. The injected-client path acquires a
+permit and invokes its client once, regardless of the internal retry budget.
+The runner continues to own the logical execution's admission lease. A request
+can override the engine's retry budget; the runner freezes that choice for
+prepared execution and passes it to each repair generation.
+
+Retry waits hold no active permit. The local delay uses equal jitter between
+half and all of a capped exponential base, beginning at 500 milliseconds and
+capping at 8 seconds. A valid provider not-before time can extend that delay;
+the caller context interrupts it. The next attempt reacquires the ordinary
+FIFO permit. Preparation failures dispatch nothing. When a positive retry
+budget ends with eligible failures, the internal exhaustion error retains the
+final cause and actual dispatched-attempt count.
+
+When an invocation-local [observation recorder](observation.md) is present in
+the context, each acquisition emits an attempt event, including canceled
+capacity waits without dispatch. Each actual retry wait emits a delay event,
+including interrupted waits. Active permits are released before callbacks;
+generation duration ends before callback delivery. Built-in successful-envelope
+usage retains reported-field presence, while injected clients use explicit
+presence when supplied and otherwise treat zero fields as unknown.
+Malformed successful envelopes can retain strictly decoded usage; error
+response bodies are never mined for it. Arbitrary injected failures carry
+unknown usage. Transport, timeout, malformed, and noneligible status failures
+stop the sequence without retry exhaustion. Context is checked before each
+wait, acquisition, and dispatch.
+
+The public client adapter translates explicitly supplied public transport and
+malformed-response identities into internal categories while retaining the
+original error chain. This keeps injected attempt classifications consistent
+with terminal observations without parsing diagnostic text or retrying injected
+clients.
+
+The diagnostic path uses the same private attempt and retry implementation as
+ordinary generation. It retains invocation-local transport, response,
+generation, usage, capacity-wait, and retry-delay evidence before normal
+result-discard and cancellation checks. Built-in preparation also supplies its
+frozen request URL to the evidence path before attempts begin; injected
+generation has no equivalent HTTP destination. This evidence is returned to
+diagnostic orchestration; it is not added to public `ExecutionEvent` values or ordinary
+observer accounting. The [model-client document](llm.md) describes the
+transport-observation boundary, and the root
+[`SmokeTestReport` GoDoc](../../smoke_diagnostics.go) owns report semantics.
+
+Focused coordinator tests use controlled transport and wait functions. They
+cover eligibility, exhaustion, cancellation, backoff, request freezing, permit
+release, and the injected-client path. The
+[evidence tests](../../internal/generation/evidence_test.go) own diagnostic
+attempt retention and retry evidence. The capacity manager tests own permit
+acquisition and idempotent release. Public
+[integration tests](../../reliable_execution_contract_test.go) cover retry,
+repair, observation, frozen prepared execution, and shared-backend behavior
+through the assembled facade.

docs/internal/llm.md

@@ -10,7 +10,7 @@ owns the observable outbound HTTP contract. The
 [framework format reference](../formats.md) owns the profile and prompt
 settings consumed by the client.
 
-The concrete client remains under `internal/llm`. The root engine assembles it
+The concrete client remains under `internal/llm`. The internal engine assembles it
 as the default implementation behind Promptkit's public client boundary.
 
 ## Components And Flow
@@ -21,32 +21,70 @@ uses internal domain values for rendered prompts, execution targets,
 structured output, responses, and token usage.
 
 The runner supplies a fully resolved target after applying backend, profile,
-and request precedence. The client uses its endpoint, credential metadata,
-generation fields, and extra parameters. `BackendID` remains routing metadata
+and request precedence, plus canonical provider-bound text messages. The
+client uses its endpoint, credential metadata, generation fields, and extra
+parameters. `BackendID` remains routing metadata
 for the generation boundary and is not mapped into the provider payload.
 
 Construction trims and validates a nonempty configured base URL and clones any
 supplied `http.Client` so Promptkit can apply its timeout default without
 mutating the caller's client. An empty configured base remains valid because a
-resolved request target may supply the endpoint. Generation then:
+resolved request target may supply the endpoint. Generation preparation then:
 
 1. validates shared execution-setting invariants and the final selected base
    endpoint;
 2. maps the internal request into the OpenAI-compatible chat payload;
-3. validates and merges extra parameters;
+3. checks extra-parameter keys against empty and reserved names, then merges
+   their values into the payload;
 4. composes `/chat/completions` through parsed URL path operations;
-5. resolves authentication;
-6. performs the outbound request under the applicable deadlines; and
-7. decodes one strictly framed, size-bounded successful response object and
-   maps its first choice and token usage, or decodes bounded structured
-   non-success detail.
+5. encodes the complete body once on success. If encoding fails, extra values
+   are checked individually to retain invalid-request classification; otherwise
+   the complete-request failure remains a request failure;
+6. resolves authentication and freezes the URL, body, and optional credential.
+
+Each execution of that prepared request creates a fresh HTTP request, body
+reader, and attempt timeout context. It closes the response body and returns
+the HTTP status, decoded response or error, reported token values and presence,
+and any valid provider `Retry-After` not-before time. A strictly decoded
+successful envelope can retain usage even if its first choice or content is
+missing. Invalid framing and non-success error bodies supply no usage. The
+ordinary `Generate` method prepares and executes once and returns no partial
+response on failure.
 
 `internal/llm` owns the set of reserved OpenAI-compatible request fields used
 when validating extra parameters. Backend registration consumes the same rule
 without making the model client depend on registry configuration.
 
-The implementation has no retry loop, tool-call support, provider catalog,
-inbound HTTP behavior, or durable session store.
+This client does not schedule retries; the [generation coordinator](generation.md)
+does so around its prepared attempts. The client has no tool-call support,
+provider catalog, inbound HTTP behavior, or durable session store.
+
+## Diagnostic HTTP Evidence
+
+Built-in diagnostics wrap the selected HTTP client's `RoundTripper` for one
+invocation using a shallow client copy, preserving its timeout, redirect policy,
+cookie jar, and underlying transport without mutating the configured client.
+Evidence distinguishes a diagnostic dispatch invocation from an observed
+request and response. A newly observed request replaces the previous
+destination-specific response evidence, so redirects retain only the last
+observed hop. A response is recorded before envelope decoding or later context
+checks can discard a generation result.
+
+The collector keeps only the last observed destination (without URL user
+information), status, credential-presence, response-availability, decoded
+generation, and presence-aware usage facts needed by the diagnostic. It does
+not retain headers, credentials, raw request or response bodies, or a redirect
+trace. These are observations at Promptkit's RoundTripper boundary, not proof
+that bytes reached a peer or that a custom transport did no hidden dispatch.
+Reachability and named authentication probes use the same configured built-in
+client but discard their response bodies after local recognition. The
+[integration contract](../integrations/openai-compatible-chat.md#explicit-smoke-diagnostics)
+owns their observable HTTP semantics.
+
+An injected `LLMClient` does not pass through this HTTP boundary. Its returned
+response can establish response and inference facts, but not an HTTP
+destination, status, reachability, or upstream credential-supply fact. The
+facade does not silently install a built-in transport for those diagnostics.
 
 ## Prepared Generation
 
@@ -89,6 +127,18 @@ truncation, malformed JSON, trailing data, and a second value are malformed
 responses with no partial result or provider content in the error. Every body
 is closed, and an unbounded oversized stream is not drained.
 
+After framing succeeds, the first choice must contain an explicitly present
+string `message.content`. The string is returned exactly, including empty or
+whitespace-only content. Missing choices, missing or `null` content, and
+non-string content are malformed responses. Output validation and correction
+eligibility remain outside this package.
+
+Successful-envelope usage is decoded with independent presence bits for each
+token field: omission and `null` are unknown, while a reported integer zero is
+known. Strictly decoded usage can still be observed if later choice/content
+checks fail; invalid framing or a non-success error body contributes none.
+Existing successful-result accounting remains value-based.
+
 For a non-success response, `ProviderHTTPError` retains the HTTP status and
 only normalized detail from the bounded recognized envelope. It retains
 `ErrUnexpectedStatus` through unwrapping. The client owns response closure;
@@ -114,7 +164,10 @@ own configuration, client cloning, deterministic deadline precedence,
 authentication, request and response mapping, malformed data, error identity,
 cancellation, endpoint selection and composition, pre-transport rejection, and
 bounded single-document successful-response framing, closure, and
-response-body suppression. The focused
+response-body suppression. The [HTTP evidence tests](../../internal/llm/http_evidence_test.go)
+own invocation-local last-hop request/response facts, while the
+[prepared-generation evidence tests](../../internal/llm/prepared_generation_evidence_test.go)
+own retention across decoding and cancellation. The focused
 [provider HTTP error tests](../../internal/llm/provider_http_error_test.go)
 own envelope parsing, normalization, and bounded-reader cases; their
 [transport tests](../../internal/llm/provider_http_error_transport_test.go)

docs/internal/observation.md

@@ -0,0 +1,56 @@
+# Internal Execution Observation
+
+`internal/observation` owns one invocation's run ID, presence-aware token
+aggregation, measured phase totals, and synchronous metadata event delivery.
+Its event values and callback type live in `internal/domain` so internal
+requests can carry them without depending on the recorder package. The
+recorder supplies run IDs to the runner and receives attempt and
+retry-delay measurements from the generation coordinator. Ordinary `Run`
+lifecycle instrumentation spans the facade and runner boundaries.
+Prepared execution starts only after its private payload is atomically claimed;
+the facade receives that invocation's recorder privately and emits the terminal
+event after public error mapping. Failed claims and preparation emit nothing.
+
+The facade masks inherited recorder and prepared-claim state when entering
+preparation or execution. Delegated calls from an injected collaborator retain
+the caller's cancellation, deadline, and unrelated context values, but do not
+reuse the enclosing invocation's accounting. Only the new request's observer
+(or the claimed handle's retained observer) enables its observation lifecycle.
+
+A recorder keeps only counters, resolved identity fields, and the selected
+callback. It has no global history, queue, goroutine, or mutex held across a
+callback. The caller supplies normalized failure metadata; the recorder never
+parses an error string or receives provider bodies, credentials, headers, or
+other content. Events are values with no mutable containers. Callback panics
+propagate to the caller.
+Resolved prompt and profile IDs are the exact runtime identities selected for
+the invocation, including namespace qualification when supplied by an engine
+view.
+Per-operation callbacks are ordered and synchronous; different operations may
+invoke the same callback concurrently. The recorder marks each measured phase's
+end before delivery, so callback work does not inflate provider duration.
+Terminal phase durations sum measured phases, while wall duration may also
+include preparation, validation, and callbacks. Panics do not promise a
+terminal event.
+
+Attempt usage is known when its explicit presence bit is true. For legacy
+injected clients without presence metadata, nonzero counts are known and
+zeros are unknown. Aggregate values sum known contributions; `Known` means
+at least one reported value, while `Complete` means every dispatched attempt
+reported the field. Zero dispatched attempts produce known, complete zero
+totals. Non-dispatched acquisitions contribute capacity wait but no usage or
+generation duration. Terminal totals count dispatched attempts, transient
+attempts above the first in a generation, and repair generations with at least
+one dispatch. A terminal event is an aggregate and must not be added to its
+attempt events.
+
+Smoke diagnostics mask inherited execution-recorder state while invoking
+generation or HTTP collaborators. Their private attempt evidence is assembled
+into a `SmokeTestReport`; it does not create ordinary run events or enter an
+enclosing run's observation totals.
+
+The [recorder tests](../../internal/observation/recorder_test.go) own usage
+presence, phase sums, event ordering, empty-run accounting, identity snapshots,
+and callback panic behavior. The [generation coordinator](generation.md) owns
+attempt and retry-delay measurements and emits their completion events after
+releasing active permits.

docs/internal/overview.md

@@ -11,25 +11,29 @@ contributor workflow and validation.
 
 | Component | Implemented responsibility | References |
 | --- | --- | --- |
-| Root `promptkit` package | Provides the supported engine facade, source, backend-registration, and injection options, public request, result, prompt-inspection, and profile-inspection values, opaque prepared-execution handles, profile construction, extension interfaces, value conversion, redacted formatting, typed capacity and generation error mapping, and engine-local profile-source assembly including application fallbacks. | [Package GoDoc](../../doc.go), [prepared execution](../../prepared_execution.go), [backend API](../../backends.go), [engine assembly](../../engine.go) |
-| `examples/go-library/prepare` | Demonstrates an offline downstream consumer using a prompt file, in-memory profile, inline input, and `Prepare`. It is not a public library package. | [Example program](../../examples/go-library/prepare/main.go) |
-| `examples/go-library/run` | Demonstrates an offline downstream consumer using a prompt file, in-memory profile, inline input, an injected deterministic model client, and `Run`. It is not a public library package. | [Example program](../../examples/go-library/run/main.go) |
-| `internal/backend` | Constructs each engine's immutable registry from the maintained built-in definitions and consumer additions, validates and defensively copies definitions through the shared JSON-value package, and consumes the LLM-owned OpenAI-compatible reserved request-field rule. | [Backend registry](../../internal/backend/registry.go) |
+| Root `promptkit` package | Provides the supported engine facade, ConfigKit source helpers, collection and resource mapping conveniences, injection options, public requests and results, prompt/profile/requirement inspection and identity inventories, concrete-request validation, explicit smoke diagnostics, observation values, opaque prepared-execution handles, value conversion, redacted formatting, and typed capacity and generation error mapping. | [Package GoDoc](../../doc.go), [observations](../../observations.go), [prepared execution](../../prepared_execution.go), [backend API](../../backends.go), [engine facade](../../engine.go), [smoke diagnostics](../../smoke_diagnostics.go), [identity listing](../../listing.go), [requirement facade](../../requirements.go) |
+| `examples/go-library/prepare` | Demonstrates offline preparation from two explicitly selected synthetic collections, source-relative imported content, a scoped view, filtered listings, and operator inventory. It is not a public library package. | [Example program](../../examples/go-library/prepare/main.go), [collection inputs](../../examples/go-library/prepare/collections/support/collection.yaml) |
+| `examples/go-library/run` | Demonstrates an offline downstream consumer using a prompt file, programmatic profile, inline input, an injected deterministic model client, and `Run`. It is not a public library package. | [Example program](../../examples/go-library/run/main.go) |
+| `examples/go-library/observe` | Demonstrates offline built-in HTTP retry and synchronous metadata observation across structured proposal and assessment runs using embedded sources, two profiles on one finite backend, and an application decision. It is not a public library package. | [Example program](../../examples/go-library/observe/main.go) |
+| `internal/backend` | Holds a defensive, immutable lookup of final backend definitions, including diagnostic metadata, plus capacity policies validated by the configuration adapter. | [Backend registry](../../internal/backend/registry.go) |
+| `internal/configuration` | Loads root and namespaced collection resources through ConfigKit; records effective field presence, validates final values including exact references and optional diagnostic metadata, captures indexed prompt definitions and source-relative message content, compiled templates, input metadata snapshots, sorted backend/profile/prompt identities, resource namespace sets, effective collection declarations, backend and presence-aware profile snapshots, and the unchanged read-only inspection view. | [Typed loader](../../internal/configuration/load.go), [identity listing](../../internal/configuration/listing.go), [internal sources](sources.md#typed-configuration-loading) |
 | `internal/capacity` | Owns engine-local bounded execution admission and FIFO model-generation permits for limited backend IDs, including cancellation-safe waiter removal and client wrapping. | [Internal capacity management](capacity.md) |
-| `internal/domain` | Defines internal framework values for requests, artifacts, prompt definitions, profiles, execution targets, rendering, generation, and validation, and owns source-neutral invariants for shared execution settings, OpenAI-compatible base endpoints, session identifiers, and output contracts. Source parsing, required fields, other source-specific normalization, defaulting, and boundary-specific error classification remain with their callers. | [Domain declarations](../../internal/domain/domain.go), [endpoint invariant](../../internal/domain/endpoint.go) |
+| `internal/domain` | Defines internal framework values for requests, artifacts, prompt definitions, requirement inspection, profiles, execution targets, rendering, generation, validation, and source-neutral smoke diagnostics, and owns source-neutral invariants for shared execution settings, OpenAI-compatible base endpoints, session identifiers, output contracts, and resource-identity syntax and qualification. Source parsing, required fields, other source-specific normalization, defaulting, and boundary-specific error classification remain with their callers. | [Domain declarations](../../internal/domain/domain.go), [requirement values](../../internal/domain/requirements.go), [endpoint invariant](../../internal/domain/endpoint.go) |
+| `internal/engine` | Assembles caller-selected configuration into one snapshot, then constructs shared capacity, rendering, validation, generation, repair, the ordinary runner, and the smoke tester for each engine. Immutable namespace views share those collaborators and qualify caller-supplied execution, inspection, and diagnostic selections without rewriting captured references. Built-in engines provide the diagnostic HTTP collaborator; injected-client engines do not. It delegates identity listing directly to the captured configuration inventory rather than the runner. | [Engine assembly](../../internal/engine/engine.go), [internal sources](sources.md#typed-configuration-loading), [internal runner](runner.md) |
+| `internal/generation` | Coordinates frozen built-in HTTP attempts and one-shot injected calls through active backend permits, including bounded status retries, cancellable backoff, ordinary attempt observations, and invocation-local evidence for diagnostics. | [Internal generation coordination](generation.md) |
 | `internal/defaults` | Defines application-neutral framework constants and constructs the default execution target. It contains no CLI, server, or inbound HTTP limits. | [Framework defaults](../../internal/defaults/defaults.go) |
 | `internal/filecatalog` | Provides deterministic YAML discovery and path helpers for operating-system filesystems and `fs.FS` sources. | [File catalog](../../internal/filecatalog/catalog.go) |
 | `internal/jsonvalue` | Validates and deeply copies bounded JSON-compatible extra-parameter and prepared-schema trees while preserving supported concrete value types and rejecting cycles or excessive depth and work. | [JSON values](../../internal/jsonvalue/jsonvalue.go) |
-| `internal/promptdef` | Loads strictly decoded, validated prompt definitions from filesystem and `fs.FS` sources, including version selection and contained file-backed message content. | [Framework formats](../formats.md), [prompt-definition repository](../../internal/promptdef/filesystem_repository.go) |
-| `internal/profile` | Loads strictly decoded, locally validated execution profiles from filesystem and `fs.FS` sources, overlays raw sources with error-preserving fallback, and resolves inherited profiles. | [Framework formats](../formats.md), [profile repositories](../../internal/profile/filesystem_repository.go), [internal sources](sources.md#profiles-and-built-ins) |
-| `internal/profile/builtin` | Embeds the built-in profile catalog, whose entries select maintained built-in backends. | [Built-in catalog](../formats.md#built-in-profile-catalog), [repository](../../internal/profile/builtin/repository.go) |
-| `internal/prompt` | Renders prompt messages from Go templates with artifact, variable, session, and cache-control data. | [Go-template renderer](../../internal/prompt/go_renderer.go) |
+| `internal/promptdef` | Provides a narrow captured-prompt lookup contract, not-found identity, and execution-ready definition hash. | [Prompt lookup](../../internal/promptdef/repository.go), [framework formats](../formats.md) |
+| `internal/profile` | Provides a narrow captured-profile lookup contract and not-found identity. | [Profile lookup](../../internal/profile/repository.go), [framework formats](../formats.md) |
+| `internal/prompt` | Renders prompt messages from Go templates with artifact, variable, session, and cache-control data, and conservatively enumerates dependencies from captured template syntax. | [Go-template renderer](../../internal/prompt/go_renderer.go), [requirement discovery](../../internal/prompt/requirements.go) |
 | `internal/artifact` | Resolves ordinary inline and unrestricted caller-selected file references into copied artifacts with metadata and hashes. | [Internal sources and validation](sources.md) |
 | `internal/validate` | Validates basic, JSON, and JSON Schema output using operating-system filesystem or `fs.FS` schema sources and creates operation-local validation plans with canonical contained schema resources. | [Framework formats](../formats.md#schemas), [internal sources and validation](sources.md) |
-| `internal/llm` | Defines the internal generation boundary and implements outbound OpenAI-compatible chat requests from resolved execution targets, including bounded structured non-success response decoding, successful-response decoding, authentication, deadline handling, and ownership of the OpenAI-compatible reserved request-field policy. | [Internal model client](llm.md) |
-| `internal/usecase` | Resolves prompt definitions and hashes, profiles, backends, and targets for exact inspection and request settings for preparation, and coordinates ordinary execution and one-attempt prepared execution across internal sources, rendering, artifact loading, operation-local validation plans, generation, capacity, and optional repair. | [Internal runner](runner.md), [prepared-execution implementation](../../internal/usecase/prepared_execution.go) |
+| `internal/llm` | Defines the internal generation boundary and implements outbound OpenAI-compatible chat requests from resolved execution targets, including bounded response decoding, authentication, deadline handling, per-invocation HTTP evidence and built-in diagnostic probes, and ownership of the OpenAI-compatible reserved request-field policy. | [Internal model client](llm.md) |
+| `internal/observation` | Owns invocation-local run IDs, presence-aware usage and phase accounting, and synchronous metadata event delivery for ordinary and claimed prepared runs and generation attempts; smoke diagnostics keep their private evidence outside ordinary observer events. | [Internal execution observation](observation.md) |
+| `internal/usecase` | Selects captured prompts and profiles, enumerates requirements from captured prompt syntax, applies request overrides and completion checks, coordinates ordinary and single-claim prepared execution across rendering, artifact loading, operation-local validation plans, generation, capacity, and bounded repair, and runs explicit smoke diagnostics with private cases, shared generation/capacity, and optional built-in HTTP probes. | [Internal runner](runner.md), [requirement inspection](../../internal/usecase/requirements_inspection.go), [prepared-execution implementation](../../internal/usecase/prepared_execution.go) |
 
-The root package assembles these internal components without exposing their
+The internal engine assembles these internal components without exposing their
 representations. Consumers depend only on the root facade.
 
 ## Maintenance

docs/internal/runner.md

@@ -4,59 +4,109 @@
 
 This document describes Promptkit's implemented internal orchestration. The
 [architecture policy](../policy/architecture.md) owns dependency and consumer
-boundaries. The [source and validation document](sources.md) owns repository,
+boundaries. The [source and validation document](sources.md) owns configuration,
 artifact, rendering, and validation behavior, while the
 [model-client document](llm.md) owns generation behavior and failure
 categories.
 
-The runner remains under `internal/usecase` and is assembled by the root
-Promptkit engine. Its concrete type is not part of the public API.
+The runner remains under `internal/usecase` and is assembled by
+`internal/engine`. Its concrete type is not part of the public API.
 The [framework format reference](../formats.md) owns prompt, profile, schema,
 and override semantics consumed by the runner.
 
 ## Collaborators
 
-`Runner` coordinates narrow internal interfaces for prompt definitions,
-profiles, backend resolution, artifacts, rendering, model generation, and
-validation. The root engine supplies one immutable registry containing the
-built-in backend and validated consumer additions, one engine-local run
-admitter, and a model client wrapped by the same capacity manager. Validation
+`Runner` coordinates narrow internal interfaces for captured prompt definitions
+and resolved profiles, artifacts, rendering, model generation, and validation.
+The internal engine supplies one engine-local run admitter and a generation
+coordinator using the same capacity manager. Validation
 plans and provider-facing schema metadata come from the validator's preparation
 interface.
-An output repairer can be injected internally, but the ordinary runner
-constructor does not enable one.
+The internal engine supplies one default output repairer through the explicit
+runner constructor, using the same coordinator as initial
+generation. The no-repair runner constructor remains available for focused
+internal callers and tests.
 
 Each invocation carries its state in request, prepared-run, and result values.
 The runner has no durable run or session store.
 
 ## Shared Prompt Selection
 
-The runner uses one prompt-selection and hashing boundary for ordinary
-preparation and exact prompt inspection. Preparation retains its early
-request-ID check before direct-session normalization; both operations then use
-the configured prompt repository to select one definition, load referenced
-message content, and calculate the same prompt hash.
+The internal engine applies each namespace view to slashless caller selections
+before they reach the runner. The runner therefore selects captured prompts and
+profiles by the resulting exact runtime IDs. Already-qualified selections pass
+through, while configured defaults and stored references are used as authored
+and are not requalified.
 
-Inspection stops after that structural lookup. It does not parse templates or
-touch profile, artifact, schema, renderer, validator, admission, or model
-collaborators. The root [`Engine.InspectPrompt`](../../engine.go) GoDoc owns
-the public operation's exact contract.
+Ordinary preparation selects the full prompt definition and construction-time
+hash. It retains its early request-ID check before direct-session
+normalization. Referenced message content and static template syntax were
+captured when the engine was built.
+
+`InspectPrompt` selects captured metadata through an optional repository
+capability, falling back to the full definition and existing hash resolution
+for internal repositories without it. It returns owned metadata without
+rendering templates or touching profile, artifact, schema, renderer, validator,
+admission, or model collaborators. The root
+[`Engine.InspectPrompt`](../../engine.go) GoDoc owns the public operation's
+exact contract.
+
+`InspectRequirements` uses the same metadata selection to resolve the
+request-profile/default identity, then obtains the captured requirement
+snapshot by the selected prompt ID. The analyzer receives only
+those identity fields rather than an execution definition. It analyzes the
+snapshot with the supplied caller variables and normalized direct-session
+bypass; it does not resolve target readiness or enter the preparation
+pipeline. Missing snapshot capability is a prompt-load failure rather than a
+source fallback. The root
+[`Engine.InspectRequirements`](../../requirements.go) GoDoc owns the public
+operation's exact contract.
 
 ## Shared Profile Selection
 
 The runner uses one profile-selection and target-resolution boundary for
 ordinary preparation and exact profile inspection. Preparation first selects a
 request profile or a prompt default; inspection begins with its required
-explicit profile ID. Both then apply the ordinary source precedence, resolve a
-named backend, and construct the effective target from framework, backend, and
-profile values.
-
-Inspection stops after the resulting endpoint and model are structurally
-validated. It does not check credential availability or perform prompt,
+explicit profile ID. Both use the captured, presence-aware target constructed
+from the final effective profile and backend configuration. Request overrides
+apply afterward; preparation checks endpoint and model completeness only after
+those overrides. Inspection can report a partial profile without requiring a
+runnable target. Neither path checks credential availability or performs prompt,
 artifact, schema, rendering, admission, or model-client work. The root
 [`Engine.InspectProfile`](../../engine.go) GoDoc owns the public operation's
 exact contract.
 
+## Smoke-Test Diagnostics
+
+The separate `SmokeTester` in `internal/usecase` selects from the same captured
+configuration snapshot and backend/profile collaborators as the runner. The
+internal engine applies namespace-view qualification to slashless caller IDs
+before diagnostic selection; captured defaults and references remain exact.
+Explicit profile selection takes precedence over prompt-default selection and a
+backend diagnostic default; HTTP-only selection does not resolve a default
+inference profile. Exact public selection rules belong to the root
+[`Engine.SmokeTest` GoDoc](../../smoke_diagnostics.go).
+
+Diagnostics do not call `Run`, `Prepare`, or the prompt preparation pipeline.
+They do not render application messages, load application artifacts, use
+application schemas, or inspect session templates. Private text and structured
+cases and the structured case's schema remain internal to the diagnostic
+operation; they are not entries in prompt/profile listings or inspection.
+`InspectConfiguration` remains a separate, explicit operator-oriented view of
+effective settings and source history. A diagnostic report may include selected
+effective settings and identifiers, so consumers still apply their disclosure
+policy.
+
+The engine assembles the smoke tester with the runner's generation coordinator
+and capacity manager. Inference holds the ordinary logical admission lease and
+uses the same active permits, retries, timeout, and request evidence path as
+generation. Reachability and authentication probes bypass generation capacity
+and do not require a model. An engine with an injected `LLMClient` can run
+inference diagnostics but has no built-in HTTP probe collaborator; HTTP-only
+diagnostics report that limitation instead of falling back to another client.
+The public GoDoc owns the caller-visible timeout, cancellation, report, and
+transport semantics.
+
 ## Shared Preparation Pipeline
 
 `Prepare` and `Run` share one private preparation pipeline split at the point
@@ -64,11 +114,10 @@ where a run can be assigned to its selected backend pool. The resolution phase
 performs only the work needed to validate routing and admission:
 
 1. validate the required prompt selection and normalize any direct session ID;
-2. load the prompt definition and hash the original definition;
+2. select the captured prompt definition and hash;
 3. select the request profile or the prompt's default profile;
-4. resolve the profile's backend ID, when present;
-5. resolve application-neutral defaults, backend defaults, profile values,
-   and explicit request overrides in that order;
+4. retain the profile's selected backend identity, when present;
+5. apply explicit request overrides to its captured backend-completed target;
 6. validate endpoint, model, numeric overrides, and credential requirements;
 7. resolve and validate the effective output contract without loading its
    schema; and
@@ -82,7 +131,8 @@ profile, or backend:
    schema metadata from it when required;
 2. load and hash input artifacts;
 3. render messages and the prompt-defined session;
-4. apply any direct session ID;
+4. apply any direct session ID, then append already-normalized request messages
+   after the rendered definition messages;
 5. hash the effective rendered prompt; and
 6. construct the prepared value and preparation timing.
 
@@ -92,30 +142,42 @@ one execution-precedence and error-ordering implementation while allowing a
 full backend pool to reject work before expensive schema, artifact, and
 rendering operations. `Prepare` discards the plan after returning its public
 metadata. `Run` retains the plan through initial and repaired-output validation
-and discards it when the operation ends. Prepared execution stores the same
-kind of plan only in its private payload.
+and discards it when the operation ends. Prepared execution transfers the
+completed preparation's owned message, input-hash, and copied-schema values
+into its private payload, while deep-copying mutable provider extras and
+creating an independent details snapshot. Claim completion or discard clears
+the execution payload without removing the details snapshot.
+
+The root facade's `Engine.ValidateRequest` uses the same complete preparation
+path as `Prepare` and discards the resulting internal prepared value without
+converting it into a public snapshot. It does not add a runner operation or
+validator component; request conversion and public error mapping remain at the
+facade boundary.
 
 Pointer-based numeric overrides preserve an explicit zero. Invalid negative or
 out-of-range values fail as invalid requests. Endpoint overrides do not change
 the selected backend identity. Non-empty extra-parameter maps replace whole
 lower-precedence maps. A direct API key takes precedence over environment
-lookup; otherwise request, profile, and backend environment-variable names
-apply in that order. A profile requiring a direct key clears an inherited
+lookup; otherwise request, explicit profile, and backend environment-variable
+names apply in that order. A profile requiring a direct key suppresses an inherited
 backend environment name unless the request supplies its own name. Secret
 values remain excluded from serialized metadata.
 
 Reasoning overrides are tri-state: nil inherits the profile, a pointer to a
 nonblank string trims and replaces it, and a pointer to a blank string clears
-it. A nonblank direct session is normalized before source loading, bypasses
+it. A nonblank direct session is normalized before prompt selection, bypasses
 the prompt session template, and is applied after ordinary message rendering.
 A blank direct value retains prompt-template behavior. The runner clears the
 template only on a value copy of the definition, so the definition hash always
-describes the original source while the rendered-prompt hash includes the
-effective direct or rendered session.
+describes captured effective content while the rendered-prompt hash includes the
+effective direct or rendered session and the complete effective message
+sequence. The rendered-prompt hash uses a versioned, length-framed SHA-256
+encoding of that session, every role and content value, and cache-control
+presence and values; its hexadecimal value is opaque.
 
-The registry is read-only after engine construction. Concurrent `Prepare` and
-`Run` calls resolve independent defensive backend values and keep all
-invocation state local.
+The configuration snapshot is read-only after engine construction. Concurrent
+`Prepare` and `Run` calls copy their resolved targets and keep all invocation
+state local.
 
 ## Run Flow
 
@@ -125,8 +187,17 @@ admitter is an internal unlimited fallback. After successful admission, `Run`
 immediately defers the returned release function, performs the completion
 phase, makes one initial generation call, builds the named output artifact,
 and validates that artifact with the plan compiled during completion. Invalid
-generated content remains a validation result; an inability to perform
-validation is an operational error.
+generated content remains a validation result; an inability to generate or
+perform validation is an operational error.
+
+When the public facade installs an invocation-local observation recorder,
+ordinary `Run` reuses its run ID and carries it to the generation coordinator
+through context. Prompt and profile identities are enriched only after their
+respective resolutions succeed, before admission. The facade starts observation
+before request conversion and emits its terminal event after public error
+mapping and after the runner's deferred admission release. Preparation and
+prepared execution follows the same terminal mapping after a successful claim;
+invalid or discarded handles do not acquire an observer.
 
 Validation preparation and execution honor cancellation at every
 Promptkit-controlled boundary and do not publish a partial plan or result.
@@ -139,20 +210,30 @@ method must return before cancellation can take precedence over its outcome.
 The admission lease covers completion-phase preparation, initial generation,
 validation, every repair, and every exit. It bounds accepted work without
 serializing preparation or validation behind the active-generation limit.
-The wrapped model client separately acquires a FIFO active permit only around
-each actual generation call.
+The coordinator separately acquires a FIFO active permit only around each
+actual model attempt. It releases the permit before a transient retry wait.
 
-When an internal repairer is present, a JSON or JSON Schema content failure can
-trigger bounded repair attempts. Repair receives the effective execution
-target, explicit numeric-presence bits, credential, backend identity, session
-ID, validation errors, prior output, and structured-output specification. One
-request constructor supplies those common fields to initial and repair
-generation while their rendered prompts remain intentionally distinct. The
-default repairer uses the same wrapped client as initial generation, so each
-repair reacquires the selected backend's active permit while remaining inside
-its original admission lease. Repair never performs a second bounded
-admission, and repaired outputs use the operation's existing validation plan.
-This capability remains internal and is not a public option.
+After a failed `basic`, JSON, or JSON Schema validation with a positive frozen
+budget, the installed repairer can make a bounded corrective call. Each request
+starts with a fresh copy of the complete effective message sequence (the
+configured prefix followed by the request suffix), includes only the latest
+nonempty candidate as an assistant message, and appends one corrective user
+message. Empty candidates omit that assistant message. The
+correction carries validation diagnostics as JSON data bounded to 64 KiB; the
+full diagnostics remain in the validation result.
+
+Repair receives the effective execution target, explicit numeric-presence bits,
+credential, backend identity, session ID, structured-output specification, and
+frozen transient retry budget.
+The same request constructor supplies those common fields to initial and repair
+generation. The default repairer uses the same coordinator as initial
+generation, so each repair reacquires the selected backend's active permit
+while remaining inside its original admission lease. Repair never performs a
+second bounded admission, and repaired outputs use the operation's existing
+validation plan. The runner stops at the first valid candidate, sums completed
+generation usage, reports calls actually started, and returns the final failed
+validation result on exhaustion. A repair generation failure follows the
+ordinary generation-error category rather than becoming a validation error.
 
 A successful result includes the output artifact and raw output, validation
 state, effective session ID, prompt and rendered-prompt hashes, selected
@@ -164,7 +245,8 @@ credential during execution, reaches generation and every repair attempt.
 Result usage is the field-wise sum of all five usage values from the initial
 response and every completed repair response. Final raw output, artifact, and
 validation state still come from the last candidate. A repair error returns no
-partial run result or partial usage.
+partial run result or partial result usage, although the invocation-local
+observer retains completed-attempt usage and reports it in the terminal event.
 
 ## Failure Categories
 
@@ -189,7 +271,7 @@ generation, validation, repair, and cancellation failures.
 Other context cancellation propagates through the invoked collaborator and is
 classified by the owning operation. In particular, cancellation observed by
 validation retains the context identity through the validation error category.
-An overlong direct session is an invalid request before source loading, while
+An overlong direct session is an invalid request before prompt selection, while
 an invalid or overlong prompt session template remains a prompt-render failure.
 An unknown selected backend, or a selected backend with no configured resolver,
 is classified as a profile-load failure.
@@ -203,11 +285,16 @@ behavior, hashing, generation and validation outcomes, backend propagation,
 bounded repair progression, initial/repair request parity, cumulative usage,
 shared initial/repair capacity, credentials and redaction, error categories,
 artifact metadata, and timing. The
-[capacity subsystem document](capacity.md) identifies the focused pool,
-waiter, and wrapped-client tests.
+[capacity subsystem document](capacity.md) identifies the focused pool and
+waiter tests. The [generation document](generation.md) owns attempt and retry
+coordination tests.
 
 Changes to orchestration should continue to use the existing package
 interfaces, keep request state local to an invocation, and preserve the shared
 resolution and completion pipeline. Source, renderer, validator, or
 model-client contract changes belong first in their owning package and
-document.
+document. The
+[smoke selection tests](../../internal/usecase/smoke_selection_test.go),
+[diagnostic orchestration tests](../../internal/usecase/smoke_tester_test.go),
+and [private inference-case tests](../../internal/usecase/smoke_inference_test.go)
+own diagnostic selection, report assembly, and inference findings.

docs/internal/sources.md

@@ -2,92 +2,163 @@
 
 ## Purpose
 
-This document describes Promptkit's implemented internal source, artifact,
+This document describes Promptkit's implemented configuration, artifact,
 rendering, and output-validation behavior. The
-[architecture policy](../policy/architecture.md) owns the library boundary and
-dependency rules. None of these internal packages is a supported consumer API,
-and the root engine assembles them behind its public source options and values.
-The [framework format reference](../formats.md) owns the exact file fields,
-validation modes, built-in catalog, and source precedence.
+[architecture policy](../policy/architecture.md) owns dependency rules, while
+the [framework format reference](../formats.md) owns fields, defaults, and
+source semantics. Internal packages are not consumer APIs.
+
+## Typed Configuration Loading
+
+`internal/configuration` is the single adapter from ConfigKit's final
+effective configuration to execution-owned definitions. It performs one
+layered load, captures effective presence (including explicit null), records
+field-level normalization and derived defaults during resolution, and validates all effective
+root and collection backend, profile, and prompt values. Replaced lower-layer
+domain values are not separately validated. It indexes root resources under
+their unqualified IDs and collection resources under their full
+`namespace/local-key` IDs while retaining ConfigKit's resolved tree. It builds an immutable backend index and presence-aware profile
+targets, applying final backend defaults only after profile inheritance. Runtime
+lookups use those final indexes and exact stored references; they do not rewrite
+or resolve resources from ConfigKit source history. The runtime/domain packages
+do not depend on ConfigKit.
+
+ConfigKit owns source headers, imports, composition, and source evidence.
+Promptkit does not preparse contributions: it validates the final effective
+values once, captures the indexed identities and their original resource
+contexts, and retains ConfigKit's resolved result for explicit operator inspection.
+Accepted non-main source headers are consumed during composition; the main
+source declaration remains available in that inspection view. Collection
+release labels are descriptive metadata, not selection inputs. The
+[format reference](../formats.md#source-headers-collections-and-resource-ids)
+owns the exact source-header and collection contracts.
+
+The `ConfigurationInspection.Diff` facade method compares two retained results
+through a private wrapper and a narrow `internal/configuration.Diff` bridge.
+The wrapper keeps `Result[document]` inside the adapter; the bridge delegates
+once to ConfigKit v1.1.0 and returns its public difference records. Comparison
+uses only frozen effective values and does not reopen sources, resolve runtime
+credentials, or enter the execution path. ConfigKit applies sensitivity from
+both results; reports may still contain ordinary application data. See the
+[consumer comparison guide](../consumers/pkg-promptkit.md#inspect-the-captured-configuration)
+and ConfigKit's [v1.1.0 comparison contract](https://forge.maximumdirect.net/base-layer/configkit/src/tag/v1.1.0/docs/config.md#effective-configuration-diffs).
+
+After resource materialization and effective-profile snapshot construction, the
+adapter captures lexically ordered identity inventories for backends, profiles,
+and prompts, per-kind resource namespace sets, and metadata for effective
+collection declarations. Listing methods return caller-owned copies from those
+captured values without consulting ConfigKit or copying resource definitions.
+
+The root `DirectoryInputs` and `FilesystemInputs` helpers discover sorted YAML
+files under an explicit root, provide resource context, and leave imports
+disabled. Each returned slice belongs to one strict-composition group; file
+order does not confer priority. Engine assembly prepends a dedicated embedded
+OpenRouter underlay unless disabled, without modifying caller-owned slices.
+Caller underlays are composed in their given
+order, followed by the optional main input. The
+[configuration tests](../../internal/configuration/load_test.go) exercise the
+real loader and final-value handoff.
+
+The public `CollectionInput` helper mounts one caller-selected source at
+`collections/<namespace>` and enables ConfigKit imports for that input. The
+helper does not read the source; ConfigKit retains responsibility for loading,
+composition, imports, and source-local resource context. Imported or inherited
+`content_file` values continue resolving relative to the source that contributed
+the effective field, not to a flattened runtime identity. Explicit
+`InspectConfiguration` views expose the retained effective configuration and
+provenance for operator review; runtime selection and inventories do not derive
+from source history.
 
 ## Prompt Definitions
 
-`internal/promptdef` uses one source-neutral flow for prompt selection and
-normalization. That flow scans normalized YAML ID and version metadata,
-requires one strictly decoded document per file, classifies errors for the
-selected definition, detects duplicates, and normalizes the exact match.
-Small operating-system and `fs.FS` adapters own discovery, byte reads, display
-paths, content opening, and root containment. Each lookup remains a
-point-in-time scan: definitions and catalogs are not cached, and file-backed
-message content is opened only for the exact selected candidate.
+The configuration adapter indexes each prompt by exact ID,
+materializes file-backed message content using the exact effective field's
+origin and resource context, and checks static message/session template syntax
+at construction. It stores captured content, compiled templates, declared
+input metadata, and a definition hash. Alongside the unchanged hashed
+definition, it retains content-type and description presence for each declared
+input and provides an internal requirement snapshot with those declarations
+and the captured parses. Metadata values come from the final normalized
+declaration; absent and null fields remain distinguishable from explicit
+strings, including empty strings. It closes content-file handles before
+returning and retains no source capability for public inspection.
 
-Operating-system sources enforce containment against canonical roots and
-targets so symlinks cannot escape. Injected `fs.FS` sources enforce containment
-in their clean relative path namespace. A single-file source uses the selected
-prompt file's containing directory as its root. Every content path must be
-relative and is opened from its exact parsed text after a separate blank check;
-contained parent components and whitespace-bearing names remain valid.
+Prompt materialization shares a lazy metadata snapshot across root and collection
+resources during construction, avoiding a full metadata copy per collection.
 
-Exact prompt inspection performs one point-in-time lookup through that same
-repository and validates referenced message content before returning declared
-metadata. It does not parse templates or read profile, input, or schema
-sources, and it does not retain the definition for a later execution.
+Normalization and defaults are recorded at individual resolved field paths,
+including message roles and cache controls. List indexes identify fields for
+presence, validation, sensitivity, and resource context; full IDs select
+captured runtime indexes. Unchanged content-file fields retain their origins.
+Backend-to-profile and request-time completion remain separate runtime work.
 
-Its package tests own prompt selection, strict decoding, definition validation,
-duplicate detection, and source containment:
-[prompt-definition repository tests](../../internal/promptdef/repository_test.go).
+The runner selects full definitions from this index for preparation and uses
+captured prompt metadata for `InspectPrompt` and `InspectRequirements`; the
+internal requirement snapshot lookup uses the selected prompt ID alone.
+These paths do not rescan configuration or reopen message files. `InspectPrompt`
+returns declared metadata without rendering or resolving profile, input, or
+schema resources, and its public shape and prompt hashes do not include the
+captured presence state. Runtime rendering still checks request variables and
+input helpers. The [prompt adapter tests](../../internal/configuration/prompts_test.go)
+cover resource origins, context containment, capture, and selection.
 
-## Profiles And Built-Ins
+The configuration adapter shares ID-only selection across its internal prompt
+lookups. The execution snapshot lookup returns an owned prompt
+definition and cached hash; metadata inspection returns owned input metadata
+without copying execution templates. Profile selection likewise has an
+identity-only lookup for consumers that do not need the execution target,
+while the full profile lookup continues to return an owned target and deep copy
+of provider extras. These lookups use captured indexes and do not query
+ConfigKit again.
 
-`internal/profile` loads, locally validates, overlays, and resolves execution
-profiles from an operating-system filesystem or an `fs.FS`. A file contains
-exactly one YAML document and its trimmed YAML `id` is its only selection
-identity; filenames do not confer authority. Each point lookup reads discovered
-files once for their metadata and reuses the selected file's bytes for strict
-decoding; unrelated profiles are not fully decoded. Strict selected decoding
-recognizes `base_profile` and the optional `backend` field, trims their values,
-and permits inherited target fields only when a base is named. File-backed
-`extra_params` values are validated and defensively copied through the shared
-bounded JSON-value owner before a profile is published. OpenAI-compatible
-reserved-field policy remains with the model-client and backend-registry owners.
+The `internal/prompt` requirement walker uses the captured parse trees and
+declaration metadata without executing templates or reading artifacts. It
+collects literal `input` names, direct root caller keys, and literal-key
+`index` lookups. For `if` and `with`, it classifies branches when their
+conditions are simple supported values: direct values, literals, negation, or
+two-string equality/inequality. It uses Go template truthiness, so a supplied
+nonempty string such as `false` is true. Missing decision values and
+content-dependent conditions leave dependency-bearing branches conditional
+and report incomplete discovery; unresolved conditions with no branch
+dependencies do not by themselves make discovery incomplete. Recognizable
+references on known skipped paths remain optional inventory entries with an
+inactive-use explanation, unless another applicable use establishes a
+stronger classification. Skipped expressions are scanned only for
+syntactically recognizable names and are not evaluated. A direct caller value
+or literal-key lookup can also supply an input name, including through simple
+parentheses or a direct pipeline to `input`; names are used exactly as
+supplied. Unresolved names and unsupported dependency-bearing expressions
+produce structured limitations while retaining recognizable references.
+Where an unresolved name or scope could affect existing optional entries, those
+entries are reported as conditional rather than inactive-only or definitively
+optional. Direct root-map `with` forms retain caller-map identity. Known scalar
+values remain separate from unknown scopes; accesses through untracked local
+maps report incomplete discovery rather than silently losing caller keys.
+Unsupported name expressions reuse value analysis's conservative syntax scan
+instead of recursively rescanning the same subtree.
+Invoked named templates are scanned conservatively, and
+dependency-bearing range scopes make discovery incomplete. Uncalled definitions
+and bypassed session templates are not traversed. This internal walker does not
+change rendering behavior.
 
-The overlay repository consults the next repository only when the
-higher-precedence repository reports that a profile is absent. A reliably
-selected malformed profile stops fallback, while an unrelated malformed file
-does not become authoritative through its filename. Loading does not check
-backend registry membership because the available registry belongs to the
-assembled engine; the runner checks membership during preparation and exact
-profile inspection.
+## Profiles
 
-The root engine assembles one raw composite catalog in precedence order:
-in-memory profiles, one ordinary configured source, an application fallback
-source, then the embedded built-in catalog. An explicit file or `fs.FS` profile
-source replaces `Config.ProfileDir` within the ordinary configured-source
-category. One outer resolving repository wraps that complete raw catalog, so
-each base lookup observes the same precedence and shadowing rules.
+ConfigKit resolves profile inheritance before the handoff. Within a collection,
+`extends: {id: parent}` names a sibling local ID; it is not a cross-collection reference.
+The configuration adapter validates final effective resource references and
+field values, then builds target
+snapshots with backend defaults and presence/suppression state. A profile can
+remain partial until runtime overrides complete it. `InspectProfile` selects
+that captured target without prompt, artifact, schema, credential lookup, or
+provider work. Preparation applies runtime overrides and then checks target
+completeness. Identity-only profile selection can identify a partial captured
+profile without copying its target or requiring readiness. The
+[target tests](../../internal/configuration/targets_test.go) cover effective
+target interpretation.
 
-The resolving repository traverses every selected chain afresh, retains no
-cache, detects cycles, limits a chain to 32 profiles, merges root-to-leaf into a
-new caller-owned value, and validates the final target before publishing it. It
-does not check backend registry membership. Exact `base_profile` syntax, merge
-rules, and consumer-visible failure behavior belong to the [framework format
-reference](../formats.md#profile-inheritance).
-
-Exact profile inspection performs one point-in-time resolved lookup through
-those profile sources and checks the final target without reading prompt, input,
-or schema sources. It does not retain that lookup for a later execution.
-Prepared execution instead freezes the fully resolved target; a later ordinary
-operation performs a fresh traversal.
-
-`internal/profile/builtin` embeds the maintained built-in profile catalog.
-Every embedded profile selects a maintained built-in backend and inherits that
-backend's endpoint and credential environment-variable name from the built-in
-backend registry rather than repeating those values. Profile loading and
-overlay behavior are owned by the
-[profile repository tests](../../internal/profile/repository_test.go), while
-catalog completeness, the backend-selection invariant, and duplicate IDs are
-owned by the
-[built-in repository tests](../../internal/profile/builtin/repository_test.go).
+The internal engine delegates listing through a narrow identity-inventory
+interface directly to the configuration snapshot. The execution runner has no
+listing responsibility.
 
 ## Ordinary Artifacts
 
@@ -98,10 +169,15 @@ and after opening the file. It copies content into an artifact, records
 metadata and an opaque content-equality value, and applies a content-type
 fallback.
 
-Regular files are read synchronously in bounded chunks. Cancellation is
-checked before opening, before and after every read, and before publishing the
-artifact, so a canceled read never publishes partial content. The ordinary
-reader does not detach file reads into background goroutines.
+Regular files are read synchronously in bounded chunks. The opened file's size
+is only an initial-capacity hint, capped at 1 MiB; reads continue to EOF, and
+the returned body determines artifact size and hash. If an overreported hint
+leaves unused capacity greater than the actual body length, the completed body
+is copied to release the excess backing storage, including when the body is
+empty. Cancellation is checked before opening, before and after every read,
+and before publishing the artifact, so a
+canceled read never publishes partial content. The ordinary reader does not
+detach file reads into background goroutines.
 
 This ordinary reader does not implement an inbound HTTP security boundary. In
 particular, it does not constrain files to an application root or impose an
@@ -112,18 +188,22 @@ implemented reader behavior and failures.
 
 ## Rendering
 
-`internal/prompt` renders definition messages as Go templates using named
-artifacts and variables. Within one render, each referenced artifact body is
-converted to text lazily and cached by input name for reuse across the session
-and every message; the cache is not shared across renders. Conversion uses
-bounded chunks and preserves the artifact bytes exactly.
+`internal/prompt` renders captured, pre-parsed definition templates using named
+artifacts and variables. It clones templates to bind request-specific input
+helpers without mutating shared state. Within one render, each referenced
+artifact body is converted to text lazily and cached by input name for reuse
+across the session and every message; the cache is not shared across renders.
+Conversion uses bounded chunks and preserves the artifact bytes exactly.
 
-Session and message parsing and execution remain synchronous. The renderer
-checks cancellation before and after each parse and execution boundary,
+Template parsing occurs during engine construction. Runtime execution remains
+synchronous. The renderer checks cancellation around execution boundaries,
 between artifact conversion chunks, around each message, and before publishing
 the complete prompt. It cannot interrupt template work already in progress and
-never publishes a partial prompt after observing cancellation. It carries
-message roles, session IDs, and cache control into the rendered prompt. The
+never publishes a partial prompt after observing cancellation. It validates and
+canonicalizes message roles before carrying roles, session IDs, and cache
+control into the rendered prompt. Message and session templates execute into
+per-call string builders, so their rendered strings do not require a final
+buffer-to-string copy. The
 [renderer tests](../../internal/prompt/renderer_test.go) own rendering behavior.
 
 ## Schemas And Output Validation
@@ -131,7 +211,8 @@ message roles, session IDs, and cache control into the rendered prompt. The
 `internal/validate` provides validators backed by an operating-system
 filesystem or an `fs.FS`. Invalid generated content is returned as a validation
 result; inability to load, register, or compile a schema is an operational
-error.
+error. Basic validation checks body emptiness on the byte slice directly
+instead of first making a response-sized string copy.
 
 Every preparation operation creates one operation-local validation plan. None,
 basic, and JSON modes retain the effective output contract without source
@@ -155,9 +236,10 @@ not detach dependency work into background goroutines.
 `Prepare` discards its validation plan after returning metadata. `Run` retains
 its plan for initial and repaired-output validation, then discards it with the
 operation. `PrepareExecution` retains the plan in its private frozen payload;
-`RunPrepared` uses that plan without reopening prompt, profile, input, or
-schema sources or rerendering the request. A later ordinary `Run` always
-performs fresh source resolution and preparation.
+`RunPrepared` uses that plan without reopening configuration, message, input,
+or schema resources or rerendering the request. A later ordinary `Run` uses
+the same captured configuration but prepares request artifacts and schemas
+afresh.
 
 The [validator tests](../../internal/validate/standard_validator_test.go) own
 basic, JSON, JSON Schema, source resolution, schema loading, compilation,

docs/oss/attribution.md

@@ -0,0 +1,118 @@
+# Open Source Attribution Notices
+
+## Standard Repository Policies
+
+Material: policy documents under `docs/policy/` and the Policy Reading Guide in
+`docs/development.md`, adapted for this repository; root `AGENTS.md`, copied
+without modification from upstream `agents/AGENTS.md`.
+
+The release policy uses upstream commit
+`e833faddf120c789a9b45348edaf75720de16348`; the remaining material uses
+`d84a38ebab26f048cafbe59a2246700c35d29843`.
+
+Copyright (C) 2026 Eric Rakestraw
+
+SPDX-License-Identifier: CC-BY-SA-4.0
+
+License and notices: [CC BY-SA 4.0](licenses/CC-BY-SA-4.0.txt) |
+[COPYRIGHT](https://gitea.maximumdirect.net/eric/policy/src/commit/d84a38ebab26f048cafbe59a2246700c35d29843/COPYRIGHT) |
+[LICENSE](https://gitea.maximumdirect.net/eric/policy/src/commit/d84a38ebab26f048cafbe59a2246700c35d29843/LICENSE) |
+[COPYRIGHT](https://gitea.maximumdirect.net/eric/policy/src/commit/e833faddf120c789a9b45348edaf75720de16348/COPYRIGHT) |
+[LICENSE](https://gitea.maximumdirect.net/eric/policy/src/commit/e833faddf120c789a9b45348edaf75720de16348/LICENSE).
+
+[Source](https://gitea.maximumdirect.net/eric/policy/src/commit/d84a38ebab26f048cafbe59a2246700c35d29843).
+
+[Source](https://gitea.maximumdirect.net/eric/policy/src/commit/e833faddf120c789a9b45348edaf75720de16348).
+
+## ConfigKit
+
+Selected module: `go.maximumdirect.net/configkit` at `v1.1.0`.
+
+Copyright (C) 2026 Eric Rakestraw
+
+SPDX-License-Identifier: LGPL-3.0-or-later
+
+License and notices: [LGPL 3.0](../../LICENSE) |
+[COPYRIGHT](https://forge.maximumdirect.net/base-layer/configkit/src/commit/191015e289d7f2b91abb9fdd9c4f2f5cf75b4de9/COPYRIGHT) |
+[LICENSE](https://forge.maximumdirect.net/base-layer/configkit/src/commit/191015e289d7f2b91abb9fdd9c4f2f5cf75b4de9/LICENSE).
+
+[Source](https://forge.maximumdirect.net/base-layer/configkit/src/commit/191015e289d7f2b91abb9fdd9c4f2f5cf75b4de9).
+
+## ReleaseKit
+
+Selected module: `go.maximumdirect.net/releasekit` at `v0.9.0` (maintainer tool).
+
+Copyright (C) 2026 Eric Rakestraw
+
+SPDX-License-Identifier: LGPL-3.0-or-later
+
+License and notices: [LGPL 3.0](../../LICENSE) |
+[COPYRIGHT](https://forge.maximumdirect.net/base-layer/releasekit/src/commit/2180aa868ec2caa95fc8279fd59d5f0ebac41bd4/COPYRIGHT) |
+[LICENSE](https://forge.maximumdirect.net/base-layer/releasekit/src/commit/2180aa868ec2caa95fc8279fd59d5f0ebac41bd4/LICENSE).
+
+[Source](https://forge.maximumdirect.net/base-layer/releasekit/src/commit/2180aa868ec2caa95fc8279fd59d5f0ebac41bd4).
+
+## jsonschema
+
+Selected module: `github.com/santhosh-tekuri/jsonschema/v6` at `v6.0.2`.
+
+SPDX-License-Identifier: Apache-2.0
+
+License and notices: [Apache 2.0](licenses/Apache-2.0.txt) |
+[LICENSE](https://github.com/santhosh-tekuri/jsonschema/blob/29cbed948d24a04700eb94436416b18a07953b71/LICENSE).
+
+[Source](https://github.com/santhosh-tekuri/jsonschema/tree/29cbed948d24a04700eb94436416b18a07953b71).
+
+## Go Text
+
+Selected module: `golang.org/x/text` at `v0.14.0` (transitive dependency).
+
+Copyright (c) 2009 The Go Authors. All rights reserved.
+
+SPDX-License-Identifier: BSD-3-Clause
+
+License and notices: [BSD 3-Clause](licenses/BSD-3-Clause.txt) |
+[LICENSE](https://github.com/golang/text/blob/6c97a165dd661335ff7bce6104a008558123c353/LICENSE) |
+[PATENTS](https://github.com/golang/text/blob/6c97a165dd661335ff7bce6104a008558123c353/PATENTS).
+
+[Source](https://github.com/golang/text/tree/6c97a165dd661335ff7bce6104a008558123c353).
+
+## Go YAML v3
+
+Selected module: `gopkg.in/yaml.v3` at `v3.0.1` (transitive dependency).
+
+Copyright (c) 2006-2010 Kirill Simonov
+
+Copyright (c) 2006-2011 Kirill Simonov
+
+Copyright (c) 2011-2019 Canonical Ltd
+
+Copyright 2011-2016 Canonical Ltd.
+
+SPDX-License-Identifier: MIT AND Apache-2.0
+
+License and notices: [MIT](licenses/MIT.txt) |
+[Apache 2.0](licenses/Apache-2.0.txt) |
+[LICENSE](https://github.com/go-yaml/yaml/blob/f6f7691b1fdeb513f56608cd2c32c51f8194bf51/LICENSE) |
+[NOTICE](https://github.com/go-yaml/yaml/blob/f6f7691b1fdeb513f56608cd2c32c51f8194bf51/NOTICE).
+
+[Source](https://github.com/go-yaml/yaml/tree/f6f7691b1fdeb513f56608cd2c32c51f8194bf51).
+
+## Go YAML v4
+
+Selected module: `go.yaml.in/yaml/v4` at `v4.0.0-rc.6` (transitive dependency).
+
+Copyright 2006-2010 Kirill Simonov
+
+Copyright 2011-2019 Canonical Ltd
+
+Copyright 2025 The go-yaml Project Contributors
+
+SPDX-License-Identifier: MIT AND Apache-2.0
+
+License and notices: [MIT](licenses/MIT.txt) |
+[Apache 2.0](licenses/Apache-2.0.txt) |
+[LICENSE](https://github.com/yaml/go-yaml/blob/1c17b9cee72bf81ae73a8be50fbd7dd3a9985125/LICENSE) |
+[NOTICE](https://github.com/yaml/go-yaml/blob/1c17b9cee72bf81ae73a8be50fbd7dd3a9985125/NOTICE).
+
+[Source](https://github.com/yaml/go-yaml/tree/1c17b9cee72bf81ae73a8be50fbd7dd3a9985125).

docs/oss/licenses/Apache-2.0.txt

@@ -0,0 +1,202 @@
+
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

docs/oss/licenses/BSD-3-Clause.txt

@@ -0,0 +1,11 @@
+Copyright (c) <year> <owner>.
+
+Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
+
+1. Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
+
+2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
+
+3. Neither the name of the copyright holder nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

docs/oss/licenses/CC-BY-SA-4.0.txt

@@ -0,0 +1,170 @@
+Creative Commons Attribution-ShareAlike 4.0 International
+
+ Creative Commons Corporation (“Creative Commons”) is not a law firm and does not provide legal services or legal advice. Distribution of Creative Commons public licenses does not create a lawyer-client or other relationship. Creative Commons makes its licenses and related information available on an “as-is” basis. Creative Commons gives no warranties regarding its licenses, any material licensed under their terms and conditions, or any related information. Creative Commons disclaims all liability for damages resulting from their use to the fullest extent possible.
+
+Using Creative Commons Public Licenses
+
+Creative Commons public licenses provide a standard set of terms and conditions that creators and other rights holders may use to share original works of authorship and other material subject to copyright and certain other rights specified in the public license below. The following considerations are for informational purposes only, are not exhaustive, and do not form part of our licenses.
+
+Considerations for licensors: Our public licenses are intended for use by those authorized to give the public permission to use material in ways otherwise restricted by copyright and certain other rights. Our licenses are irrevocable. Licensors should read and understand the terms and conditions of the license they choose before applying it. Licensors should also secure all rights necessary before applying our licenses so that the public can reuse the material as expected. Licensors should clearly mark any material not subject to the license. This includes other CC-licensed material, or material used under an exception or limitation to copyright. More considerations for licensors.
+
+Considerations for the public: By using one of our public licenses, a licensor grants the public permission to use the licensed material under specified terms and conditions. If the licensor’s permission is not necessary for any reason–for example, because of any applicable exception or limitation to copyright–then that use is not regulated by the license. Our licenses grant only permissions under copyright and certain other rights that a licensor has authority to grant. Use of the licensed material may still be restricted for other reasons, including because others have copyright or other rights in the material. A licensor may make special requests, such as asking that all changes be marked or described.
+
+Although not required by our licenses, you are encouraged to respect those requests where reasonable. More considerations for the public.
+
+Creative Commons Attribution-ShareAlike 4.0 International Public License
+
+By exercising the Licensed Rights (defined below), You accept and agree to be bound by the terms and conditions of this Creative Commons Attribution-ShareAlike 4.0 International Public License ("Public License"). To the extent this Public License may be interpreted as a contract, You are granted the Licensed Rights in consideration of Your acceptance of these terms and conditions, and the Licensor grants You such rights in consideration of benefits the Licensor receives from making the Licensed Material available under these terms and conditions.
+
+Section 1 – Definitions.
+
+     a.	Adapted Material means material subject to Copyright and Similar Rights that is derived from or based upon the Licensed Material and in which the Licensed Material is translated, altered, arranged, transformed, or otherwise modified in a manner requiring permission under the Copyright and Similar Rights held by the Licensor. For purposes of this Public License, where the Licensed Material is a musical work, performance, or sound recording, Adapted Material is always produced where the Licensed Material is synched in timed relation with a moving image.
+
+     b.	Adapter's License means the license You apply to Your Copyright and Similar Rights in Your contributions to Adapted Material in accordance with the terms and conditions of this Public License.
+
+     c.	BY-SA Compatible License means a license listed at creativecommons.org/compatiblelicenses, approved by Creative Commons as essentially the equivalent of this Public License.
+
+     d.	Copyright and Similar Rights means copyright and/or similar rights closely related to copyright including, without limitation, performance, broadcast, sound recording, and Sui Generis Database Rights, without regard to how the rights are labeled or categorized. For purposes of this Public License, the rights specified in Section 2(b)(1)-(2) are not Copyright and Similar Rights.
+
+     e.	Effective Technological Measures means those measures that, in the absence of proper authority, may not be circumvented under laws fulfilling obligations under Article 11 of the WIPO Copyright Treaty adopted on December 20, 1996, and/or similar international agreements.
+
+     f.	Exceptions and Limitations means fair use, fair dealing, and/or any other exception or limitation to Copyright and Similar Rights that applies to Your use of the Licensed Material.
+
+     g.	License Elements means the license attributes listed in the name of a Creative Commons Public License. The License Elements of this Public License are Attribution and ShareAlike.
+
+     h.	Licensed Material means the artistic or literary work, database, or other material to which the Licensor applied this Public License.
+
+     i.	Licensed Rights means the rights granted to You subject to the terms and conditions of this Public License, which are limited to all Copyright and Similar Rights that apply to Your use of the Licensed Material and that the Licensor has authority to license.
+
+     j.	Licensor means the individual(s) or entity(ies) granting rights under this Public License.
+
+     k.	Share means to provide material to the public by any means or process that requires permission under the Licensed Rights, such as reproduction, public display, public performance, distribution, dissemination, communication, or importation, and to make material available to the public including in ways that members of the public may access the material from a place and at a time individually chosen by them.
+
+     l.	Sui Generis Database Rights means rights other than copyright resulting from Directive 96/9/EC of the European Parliament and of the Council of 11 March 1996 on the legal protection of databases, as amended and/or succeeded, as well as other essentially equivalent rights anywhere in the world.
+
+     m.	You means the individual or entity exercising the Licensed Rights under this Public License. Your has a corresponding meaning.
+
+Section 2 – Scope.
+
+     a.	License grant.
+
+          1. Subject to the terms and conditions of this Public License, the Licensor hereby grants You a worldwide, royalty-free, non-sublicensable, non-exclusive, irrevocable license to exercise the Licensed Rights in the Licensed Material to:
+
+               A. reproduce and Share the Licensed Material, in whole or in part; and
+
+               B. produce, reproduce, and Share Adapted Material.
+
+          2. Exceptions and Limitations. For the avoidance of doubt, where Exceptions and Limitations apply to Your use, this Public License does not apply, and You do not need to comply with its terms and conditions.
+
+          3. Term. The term of this Public License is specified in Section 6(a).
+
+          4. Media and formats; technical modifications allowed. The Licensor authorizes You to exercise the Licensed Rights in all media and formats whether now known or hereafter created, and to make technical modifications necessary to do so. The Licensor waives and/or agrees not to assert any right or authority to forbid You from making technical modifications necessary to exercise the Licensed Rights, including technical modifications necessary to circumvent Effective Technological Measures. For purposes of this Public License, simply making modifications authorized by this Section 2(a)(4) never produces Adapted Material.
+
+          5. Downstream recipients.
+
+               A. Offer from the Licensor – Licensed Material. Every recipient of the Licensed Material automatically receives an offer from the Licensor to exercise the Licensed Rights under the terms and conditions of this Public License.
+
+               B. Additional offer from the Licensor – Adapted Material. Every recipient of Adapted Material from You automatically receives an offer from the Licensor to exercise the Licensed Rights in the Adapted Material under the conditions of the Adapter’s License You apply.
+
+               C. No downstream restrictions. You may not offer or impose any additional or different terms or conditions on, or apply any Effective Technological Measures to, the Licensed Material if doing so restricts exercise of the Licensed Rights by any recipient of the Licensed Material.
+
+          6. No endorsement. Nothing in this Public License constitutes or may be construed as permission to assert or imply that You are, or that Your use of the Licensed Material is, connected with, or sponsored, endorsed, or granted official status by, the Licensor or others designated to receive attribution as provided in Section 3(a)(1)(A)(i).
+
+     b.	Other rights.
+
+          1. Moral rights, such as the right of integrity, are not licensed under this Public License, nor are publicity, privacy, and/or other similar personality rights; however, to the extent possible, the Licensor waives and/or agrees not to assert any such rights held by the Licensor to the limited extent necessary to allow You to exercise the Licensed Rights, but not otherwise.
+
+          2. Patent and trademark rights are not licensed under this Public License.
+
+          3. To the extent possible, the Licensor waives any right to collect royalties from You for the exercise of the Licensed Rights, whether directly or through a collecting society under any voluntary or waivable statutory or compulsory licensing scheme. In all other cases the Licensor expressly reserves any right to collect such royalties.
+
+Section 3 – License Conditions.
+
+Your exercise of the Licensed Rights is expressly made subject to the following conditions.
+
+     a.	Attribution.
+
+          1. If You Share the Licensed Material (including in modified form), You must:
+
+               A. retain the following if it is supplied by the Licensor with the Licensed Material:
+
+                    i.	identification of the creator(s) of the Licensed Material and any others designated to receive attribution, in any reasonable manner requested by the Licensor (including by pseudonym if designated);
+
+                    ii.	a copyright notice;
+
+                    iii. a notice that refers to this Public License;
+
+                    iv.	a notice that refers to the disclaimer of warranties;
+
+                    v.	a URI or hyperlink to the Licensed Material to the extent reasonably practicable;
+
+               B. indicate if You modified the Licensed Material and retain an indication of any previous modifications; and
+
+               C. indicate the Licensed Material is licensed under this Public License, and include the text of, or the URI or hyperlink to, this Public License.
+
+          2. You may satisfy the conditions in Section 3(a)(1) in any reasonable manner based on the medium, means, and context in which You Share the Licensed Material. For example, it may be reasonable to satisfy the conditions by providing a URI or hyperlink to a resource that includes the required information.
+
+          3. If requested by the Licensor, You must remove any of the information required by Section 3(a)(1)(A) to the extent reasonably practicable.
+
+     b.	ShareAlike.In addition to the conditions in Section 3(a), if You Share Adapted Material You produce, the following conditions also apply.
+
+          1. The Adapter’s License You apply must be a Creative Commons license with the same License Elements, this version or later, or a BY-SA Compatible License.
+
+          2. You must include the text of, or the URI or hyperlink to, the Adapter's License You apply. You may satisfy this condition in any reasonable manner based on the medium, means, and context in which You Share Adapted Material.
+
+          3. You may not offer or impose any additional or different terms or conditions on, or apply any Effective Technological Measures to, Adapted Material that restrict exercise of the rights granted under the Adapter's License You apply.
+
+Section 4 – Sui Generis Database Rights.
+
+Where the Licensed Rights include Sui Generis Database Rights that apply to Your use of the Licensed Material:
+
+     a.	for the avoidance of doubt, Section 2(a)(1) grants You the right to extract, reuse, reproduce, and Share all or a substantial portion of the contents of the database;
+
+     b.	if You include all or a substantial portion of the database contents in a database in which You have Sui Generis Database Rights, then the database in which You have Sui Generis Database Rights (but not its individual contents) is Adapted Material, including for purposes of Section 3(b); and
+
+     c.	You must comply with the conditions in Section 3(a) if You Share all or a substantial portion of the contents of the database.
+For the avoidance of doubt, this Section 4 supplements and does not replace Your obligations under this Public License where the Licensed Rights include other Copyright and Similar Rights.
+
+Section 5 – Disclaimer of Warranties and Limitation of Liability.
+
+     a.	Unless otherwise separately undertaken by the Licensor, to the extent possible, the Licensor offers the Licensed Material as-is and as-available, and makes no representations or warranties of any kind concerning the Licensed Material, whether express, implied, statutory, or other. This includes, without limitation, warranties of title, merchantability, fitness for a particular purpose, non-infringement, absence of latent or other defects, accuracy, or the presence or absence of errors, whether or not known or discoverable. Where disclaimers of warranties are not allowed in full or in part, this disclaimer may not apply to You.
+
+     b.	To the extent possible, in no event will the Licensor be liable to You on any legal theory (including, without limitation, negligence) or otherwise for any direct, special, indirect, incidental, consequential, punitive, exemplary, or other losses, costs, expenses, or damages arising out of this Public License or use of the Licensed Material, even if the Licensor has been advised of the possibility of such losses, costs, expenses, or damages. Where a limitation of liability is not allowed in full or in part, this limitation may not apply to You.
+
+     c.	The disclaimer of warranties and limitation of liability provided above shall be interpreted in a manner that, to the extent possible, most closely approximates an absolute disclaimer and waiver of all liability.
+
+Section 6 – Term and Termination.
+
+     a.	This Public License applies for the term of the Copyright and Similar Rights licensed here. However, if You fail to comply with this Public License, then Your rights under this Public License terminate automatically.
+
+     b.	Where Your right to use the Licensed Material has terminated under Section 6(a), it reinstates:
+
+          1. automatically as of the date the violation is cured, provided it is cured within 30 days of Your discovery of the violation; or
+
+          2. upon express reinstatement by the Licensor.
+
+     c.	For the avoidance of doubt, this Section 6(b) does not affect any right the Licensor may have to seek remedies for Your violations of this Public License.
+
+     d.	For the avoidance of doubt, the Licensor may also offer the Licensed Material under separate terms or conditions or stop distributing the Licensed Material at any time; however, doing so will not terminate this Public License.
+
+     e.	Sections 1, 5, 6, 7, and 8 survive termination of this Public License.
+
+Section 7 – Other Terms and Conditions.
+
+     a.	The Licensor shall not be bound by any additional or different terms or conditions communicated by You unless expressly agreed.
+
+     b.	Any arrangements, understandings, or agreements regarding the Licensed Material not stated herein are separate from and independent of the terms and conditions of this Public License.
+
+Section 8 – Interpretation.
+
+     a.	For the avoidance of doubt, this Public License does not, and shall not be interpreted to, reduce, limit, restrict, or impose conditions on any use of the Licensed Material that could lawfully be made without permission under this Public License.
+
+     b.	To the extent possible, if any provision of this Public License is deemed unenforceable, it shall be automatically reformed to the minimum extent necessary to make it enforceable. If the provision cannot be reformed, it shall be severed from this Public License without affecting the enforceability of the remaining terms and conditions.
+
+     c.	No term or condition of this Public License will be waived and no failure to comply consented to unless expressly agreed to by the Licensor.
+
+     d.	Nothing in this Public License constitutes or may be interpreted as a limitation upon, or waiver of, any privileges and immunities that apply to the Licensor or You, including from the legal processes of any jurisdiction or authority.
+
+Creative Commons is not a party to its public licenses. Notwithstanding, Creative Commons may elect to apply one of its public licenses to material it publishes and in those instances will be considered the “Licensor.” Except for the limited purpose of indicating that material is shared under a Creative Commons public license or as otherwise permitted by the Creative Commons policies published at creativecommons.org/policies, Creative Commons does not authorize the use of the trademark “Creative Commons” or any other trademark or logo of Creative Commons without its prior written consent including, without limitation, in connection with any unauthorized modifications to any of its public licenses or any other arrangements, understandings, or agreements concerning use of licensed material. For the avoidance of doubt, this paragraph does not form part of the public licenses.
+
+Creative Commons may be contacted at creativecommons.org.

docs/oss/licenses/MIT.txt

@@ -0,0 +1,18 @@
+MIT License
+
+Copyright (c) <year> <copyright holders>
+
+Permission is hereby granted, free of charge, to any person obtaining a copy of this software and
+associated documentation files (the "Software"), to deal in the Software without restriction, including
+without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the
+following conditions:
+
+The above copyright notice and this permission notice shall be included in all copies or substantial
+portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT
+LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO
+EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER
+IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+USE OR OTHER DEALINGS IN THE SOFTWARE.

docs/policy/agents.md

@@ -0,0 +1,96 @@
+# Agent Policy
+
+**Base Policy Version:** 1.3
+
+**Last Updated**: 2026-09-27
+
+**Upstream Source:** [agents.md](https://gitea.maximumdirect.net/eric/policy/src/commit/d84a38ebab26f048cafbe59a2246700c35d29843/docs/policy/agents.md)
+
+## Purpose
+
+Agents must read and follow these rules.
+
+## Implementation Discipline
+
+**KEY PRINCIPLES:** Write the simplest idiomatic code that satisfies the user's
+request. Do not implicitly strengthen requirements, restrictions, or guarantees
+beyond what the user actually requested.
+
+**Keep implementation machinery proportionate.** Use straightforward,
+idiomatic code and introduce only the machinery needed to support the requested
+behavior and flexibility. Optimize for conceptual simplicity and
+maintainability, not for the fewest lines, fewest options, or narrowest product
+surface.
+
+**Do not invent restrictions.** Do not reject, prohibit, constrain, cap,
+harden, or otherwise narrow behavior unless required by the user, an explicit
+repository contract, or an unavoidable correctness constraint. Preserve
+operator and caller choice where policy has not been established. If an
+additional restriction appears advisable, present it as a decision rather than
+silently implementing it.
+
+**Do not invent guarantees.** Do not strengthen public contracts, guarantees,
+or invariants beyond those expressly requested by the user.
+
+**Use the standard library.** For example: in almost every case, the best
+approach to copying files is to use the standard library. Writing custom code to
+implement power-loss protection at the application level, or to guard against
+bad actors with privileged and concurrent access to the host's filesystem,
+is wasteful, unnecessary, and not idiomatic. Stick to the simple, standard, and
+idiomatic approach unless another path is expressly requested by the user.
+
+**An implementation detail is not repository policy.** Tests and documentation
+must not elevate incidental behavior into a durable guarantee or restriction
+unless that behavior was expressly selected as part of the contract.
+
+## Promptkit Contract Authority
+
+The following clarification preserves Promptkit's established distinction
+between implementation choices and authorized contracts.
+
+A correctness constraint must follow from the requested behavior or an
+established invariant, not from the convenience of the chosen implementation.
+Explain the specific failure it prevents and why a less restrictive
+implementation cannot satisfy the contract.
+
+Existing explicit contracts remain authoritative; changing or weakening one
+requires a deliberate decision. Agent-authored plans, tests, documentation,
+and prior implementations are not evidence that an additional restriction was
+deliberately approved. Identify the authorizing requirement or decision; when
+that authority is unclear, surface the ambiguity rather than treating repetition
+as approval.
+
+## Git Commit Policy
+
+By default, commit each completed, nonempty unit of work using a
+concise, plain-English message describing what changed. This policy is standing
+authorization to create those local commits; a separate user request is not
+required. An explicit user instruction to leave work uncommitted takes precedence.
+
+A unit of work is a focused, reviewable outcome, including its relevant tests and
+documentation.
+Complete the appropriate validation before committing. If work is incomplete or
+validation is blocked, report that state accurately.
+
+Before committing:
+
+1. Run `git status --short` and review the relevant diff, including any staged
+   changes. Identify changes that predate the current unit of work.
+2. Exclude secrets, local environment files, generated junk, and unrelated changes.
+3. Stage only the files or hunks belonging to the completed unit. Use `git add .`
+   only when every included change belongs to that unit. Preserve unrelated
+   staged and unstaged work.
+4. Review `git diff --staged` and run appropriate checks for the change. Reuse
+   successful checks for unchanged content.
+5. Review `git log --oneline -n 10` and use
+   `git commit -m "<concise plain-English commit message>"`.
+6. Report the commit identifier and any remaining uncommitted work.
+
+If the unit cannot be safely separated from pre-existing work, leave it uncommitted
+and explain the overlap rather than silently including that work. If a commit
+fails, preserve the changes and report the failure; do not bypass hooks.
+
+Default commit authorization does not authorize pushing, publishing, amending,
+resetting, reverting, or rewriting history. Those actions require a separate user
+instruction. Do not use `git commit -a` or `git commit --no-verify` to bypass the
+staging and validation discipline above.

docs/policy/architecture.md

@@ -1,11 +1,41 @@
 # Architecture Policy
 
-## Purpose
+**Base Policy Version:** 1.3
 
-This document defines Promptkit's current high-level architecture and the
-durable boundaries that implementation changes must preserve. The
-[internal component overview](../internal/overview.md) inventories concrete
-implemented packages without redefining these rules.
+**Last Updated**: 2026-09-27
+
+**Upstream Source:** [architecture.md](https://gitea.maximumdirect.net/eric/policy/src/commit/d84a38ebab26f048cafbe59a2246700c35d29843/docs/policy/architecture.md)
+
+## Purpose And Ownership
+
+The adopted policy defines Promptkit's purpose and the architectural
+constraints contributors must follow. It owns system boundaries, responsibility
+and dependency rules, established invariants, and explicit non-goals.
+
+Describe the concrete package inventory and implementation mechanics in internal
+documentation. Link to canonical public contracts for their exact behavior and
+to ADRs for decision rationale.
+
+## Architectural Approach
+
+Hexagonal architecture is strongly recommended, including encapsulating external
+integrations and APIs within internal adapters. Keep domain and application
+logic independent of external integration details, and make dependency direction
+clear in the adopted policy.
+
+Implement that separation with proportionate, idiomatic code. Choose interfaces
+and package layout to serve the application's actual boundaries. Use adapters
+for collaborators with external integration responsibilities.
+
+## Adopted Constraints And Implemented Structure
+
+Architectural constraints govern current and future development. Record
+implemented structure accurately, proposals and implementation status in the
+roadmap, and significant decisions and rationale in ADRs. Describe a target
+structure as implemented when its changes land.
+
+Constraints must reflect deliberately selected requirements and decisions.
+Do not promote incidental implementation choices into architectural invariants.
 
 ## System Shape
 
@@ -17,55 +47,26 @@ The module root contains package `promptkit`, which is the public facade. It
 provides the supported engine, configuration and source options, requests,
 results, public values, extension interfaces, profiles, and error sentinels.
 
-The implemented internal components consist of:
+The [internal component overview](../internal/overview.md) owns the inventory
+of implemented packages and their concrete responsibilities. This policy owns
+the boundaries and dependency rules those components must preserve.
 
-- `internal/domain`, which owns framework data values and source-neutral
-  invariants shared by later internal components;
-- `internal/backend`, which owns validated immutable OpenAI-compatible backend
-  definitions and the maintained built-in definitions;
-- `internal/capacity`, which owns engine-local bounded run admission and
-  model-generation scheduling for limited backends;
-- `internal/defaults`, which owns application-neutral framework defaults and
-  constructs the default execution target;
-- `internal/filecatalog`, which discovers YAML files and provides source-path
-  helpers for filesystem and `fs.FS` consumers;
-- `internal/jsonvalue`, which validates and defensively copies JSON-compatible
-  extra-parameter trees;
-- `internal/promptdef`, which loads and validates prompt definitions from
-  filesystem and `fs.FS` sources;
-- `internal/profile`, which loads, validates, and overlays execution profiles
-  from filesystem and `fs.FS` sources;
-- `internal/profile/builtin`, which embeds the built-in execution profile
-  catalog;
-- `internal/prompt`, which renders prompt messages from Go templates;
-- `internal/artifact`, which resolves ordinary inline and unrestricted
-  caller-selected file references;
-- `internal/validate`, which validates basic, JSON, and JSON Schema output
-  using filesystem and `fs.FS` schema sources;
-- `internal/llm`, which defines the provider-neutral generation boundary and
-  implements outbound OpenAI-compatible chat requests; and
-- `internal/usecase`, which coordinates preparation and execution across the
-  internal framework components.
+The internal engine composes caller-selected ConfigKit inputs once, then
+assembles the captured definitions, capacity manager, renderer, validator,
+outbound client, and use-case runner. The configuration adapter interprets and
+validates only the final effective values, including presence, null, and
+resource context. It retains source history for explicit operator inspection,
+not for validation or a second resolution pass. The root facade translates
+public values and errors at the library boundary.
 
-The `examples/go-library/prepare` and `examples/go-library/run` packages are
-maintained downstream consumers of the root facade. They do not expose library
-packages or participate in internal assembly.
-
-The root facade assembles one immutable backend registry, one capacity manager,
-the internal repositories, renderer, validator, outbound client, and use-case
-runner while translating public values and errors at the library boundary. The
-registry contains built-ins plus validated engine-scoped consumer additions.
-The facade constructs the capacity manager from the registry's immutable
-policy snapshot, wraps the selected built-in or injected model client, and
-supplies bounded admission to the runner. The defaults and renderer depend on
-the domain model. Prompt-definition and profile repositories use the domain
-model, file catalog, and YAML decoder. The built-in profile repository supplies
-an embedded `fs.FS` to the profile package. Artifact reading uses the domain
-model and application-neutral defaults. Validation uses the domain model, file
-catalog, and JSON Schema implementation. The model client uses the domain
-model, application-neutral defaults, and an injected or standard-library HTTP
-client. The use-case runner depends on the narrow interfaces owned by each
-internal component, including backend lookup and run admission.
+The engine constructs capacity policy from final backend definitions, wraps
+the selected built-in or injected model client, and supplies bounded admission
+to the runner. Runtime/domain components consume execution-owned values, not
+ConfigKit types. Artifact reading and validation use the domain model and
+application-neutral defaults; validation also uses file-catalog helpers and
+JSON Schema. The model client uses the domain model, defaults, and an injected
+or standard-library HTTP client. The runner coordinates narrow lookup and
+collaborator interfaces without rereading configuration.
 
 The current implementation follows this dependency direction:
 
@@ -82,9 +83,10 @@ downstream consumers, including Scriptorium
         narrow injected abstractions
 ```
 
-The backend registry depends on the domain model and shared JSON-value
-validation, has no mutation API after construction, and consumes the
-OpenAI-compatible reserved request-field rule owned by the model client. The
+The backend index depends on the domain model and shared JSON-value ownership
+and has no mutation API after construction. The configuration adapter applies
+the OpenAI-compatible reserved request-field rule owned by the model client
+when it validates final provider maps. The
 capacity component depends on the domain model and the narrow internal
 model-client boundary, not on provider transport implementation. The model
 client does not depend on registry or capacity configuration. The facade
@@ -95,7 +97,8 @@ on consumers or on Scriptorium.
 `internal/domain` owns source-neutral invariants for values shared across
 multiple input and execution boundaries, including execution-setting bounds,
 OpenAI-compatible base endpoints, session identifiers, and output-contract
-legality. Callers retain source parsing, required-field rules, other
+legality, plus shared resource-identity syntax and qualification. Callers retain
+source parsing, required-field rules, other
 source-specific normalization, defaulting, error classification, and policy
 specific to their own boundary.
 
@@ -125,8 +128,10 @@ consumer contract requires a public package.
 Each package must have one cohesive responsibility and a clear dependency
 direction. Internal packages must not expose their types merely to simplify
 wiring, and the public facade must not leak internal representations through
-exported signatures. New public packages require a durable consumer need that
-cannot be served cleanly by the root facade.
+exported signatures. Production packages under `internal/` must not import the
+module root; the root facade may depend on internal implementation packages.
+New public packages require a durable consumer need that cannot be served
+cleanly by the root facade.
 
 The [internal component overview](../internal/overview.md) must be updated as
 packages are implemented or their responsibilities change.
@@ -177,3 +182,21 @@ state.
 Do not list planned packages as implemented components. When implementation
 introduces a package, update the internal inventory and the owning contract or
 subsystem document in the same change.
+
+## Trust Boundaries And Threat Scope
+
+Where useful, state the trust boundaries and threat scenarios relevant to the
+application's adopted requirements, along with scenarios it explicitly excludes.
+For example, distinguish handling untrusted application input from defending
+against a privileged actor concurrently modifying the host filesystem.
+
+Threat exclusions define the scope of the adopted requirements. Base defenses
+and guarantees on those requirements and raise unsettled scope for a deliberate
+decision under the [agent policy](agents.md).
+
+Keep this discussion proportionate to the application. A separate threat
+assessment or security document is optional when adopting this template.
+For this library, the framework format and integration documents own credential
+supply and external contracts; internal documents own implementation mechanisms.
+Consumer-specific filesystem and security policy remains outside Promptkit's
+boundary as described above.

docs/policy/documentation.md

@@ -1,138 +1,139 @@
 # Documentation Policy
 
+**Base Policy Version:** 1.11
+
+**Last Updated**: 2026-09-27
+
+**Upstream Source:** [documentation.md](https://gitea.maximumdirect.net/eric/policy/src/commit/d84a38ebab26f048cafbe59a2246700c35d29843/docs/policy/documentation.md)
+
 ## Purpose
 
-This policy assigns each Promptkit documentation topic to one canonical owner.
-Its goal is to keep documentation for this reusable Go library accurate,
-concise, discoverable, and resistant to drift for consumers, contributors,
-maintainers, integrators, and coding agents.
+This policy assigns each documentation topic to one canonical owner. Its goal is
+to keep this library's documentation accurate, concise, discoverable, and resistant
+to drift for users, operators, developers, integrators, and LLM coding agents.
 
 ## Core Rules
 
 ### One Canonical Owner
 
-Each authoritative fact belongs in one document or source form. A non-owning
-document may give a short, stable summary for orientation, but it must link to
-the canonical owner instead of repeating exact contracts.
+Each authoritative fact belongs in one document. A non-owning document may give
+a short, stable summary for orientation, but it must link to the canonical owner
+for volatile details.
 
-Volatile details include exported declarations, accepted inputs, defaults,
-schemas, file names, paths, error identities, retry behavior, and runtime
-guarantees. If readers could reasonably treat a statement as a contract,
-maintain its exact definition only in the owning source.
+Volatile details include commands, flags, configuration fields and defaults,
+module keys, schemas, file names, paths, status codes, retry behavior, and
+runtime guarantees. If readers could reasonably treat a statement as a
+contract, maintain it only in the owning document.
 
-### Current State, Decisions, And Future Work
+### Current And Future Behavior
 
 Outside `docs/roadmap/`, documentation describes implemented behavior only.
 Partial features may be described only to their implemented boundary.
 
-An accepted architecture decision may describe an approved direction before it
-is implemented, but acceptance is not evidence that the behavior exists.
-Current-state documents change when the implementation lands. Temporary
-roadmaps own future work, sequencing, and implementation status; they do not
-replace durable policies or current contracts.
+Architectural constraints may govern future changes. Describe current structure
+from implemented behavior.
+
+ADRs may record an accepted architectural decision before implementation. The
+roadmap owns implementation status and sequencing until the decision is
+implemented. Current architecture, user, operator, integration, and internal
+documentation are updated when the behavior lands.
 
 ### Audience And Detail
 
 Write for the document's stated audience and include only the detail needed for
-its owned topic. Consumer guidance should not expose incidental implementation
-detail. Contributor documentation should link to public contracts and durable
-policies instead of restating them.
+its owned topic. Keep user and operator docs focused on using and operating the
+application. Developer docs should link to user-facing and external contracts
+for their definitions.
 
-### Links
+### Examples
 
-Use descriptive link text and repository-relative links for repository
-documents. Link to the canonical owner rather than to a duplicate summary.
-Check every added or changed link and repair or remove links when their target
-moves or is retired.
+Complete copyable files belong in `examples/`. Documentation may use the
+smallest illustrative snippet needed to explain its owned topic, but should link
+to maintained examples for complete copyable files.
 
-### Examples And Code Fences
-
-Complete copyable files belong in `examples/` when maintained examples exist.
-Documentation may use the smallest illustrative snippet needed for its owned
-topic, but should link to a maintained example instead of embedding a second
-complete copy.
-
-Examples must be valid, secret-free, and tested where practical. Commands,
-imports, and Go snippets must match the implemented library. Use a language tag
-on fenced code blocks and make clear when a fragment is illustrative rather
-than directly runnable.
+Examples must be valid, secret-free, and tested where practical. Commands and
+configuration used in documentation should match the application.
 
 ### Security And Privacy
 
 Documentation and examples must not contain real credentials, private keys,
-private environment dumps, sensitive source material, or private
-infrastructure details unless intentionally public. Document secret-handling
-mechanisms, not secret values.
+private environment dumps, sensitive source material, or private infrastructure
+details unless intentionally public. Document secret-handling mechanisms, not
+secret values.
 
 ## Canonical Ownership
 
 | Topic | Canonical owner | Owned content | Content owned elsewhere |
 | --- | --- | --- | --- |
-| Project orientation | `README.md` | What Promptkit is, its current usability, module identity, license summary, and links onward. | Exact API contracts, contributor procedures, architecture detail, and release steps. |
-| Contributor workflow | `docs/development.md` | Task-oriented reading guide, local workflow, validation commands, and repository hygiene. | Architecture rules, API semantics, subsystem behavior, and release procedure. |
-| Current architecture | `docs/policy/architecture.md` | System shape, normative ownership, dependency direction, package boundaries, invariants, and non-goals. | Concrete component inventory, implementation mechanics, contributor procedures, decision history, and future work. |
-| Documentation organization | `docs/policy/documentation.md` | Documentation ownership, audience boundaries, maintenance rules, and document lifecycle. | Library architecture or runtime behavior. |
-| Testing policy | `docs/policy/testing.md` | Test philosophy, risk-based sufficiency, test boundaries, doubles, coverage guidance, regression policy, and test maintenance. | Subsystem behavior, exact public contracts, subsystem-specific test inventories, and implementation plans. |
-| Release procedure | `docs/release.md`, when present | Required release validation, version and tag procedure, release ordering, and post-publication checks. | General contributor workflow, public API semantics, and decision history. |
+| Product orientation and minimal end-to-end quickstart | `README.md` | Purpose, value, shortest successful invocation, and navigation, including licensing and attribution. | Complete command reference, configuration reference, operational procedures, implementation detail. |
+| Contributor entry point | `docs/development.md` | Task-oriented reading guide, contributor orientation, baseline validation commands, and canonical links. | Package inventory, architecture rules, subsystem behavior, detailed change recipes. |
+| Application architecture policy | `docs/policy/architecture.md` | System shape, normative ownership, dependency direction, boundaries, invariants, safety properties, and non-goals. | Concrete package inventory, implementation mechanics, contributor procedures, decision history, future work. |
+| Policy-set provenance and upstream navigation | `docs/policy/meta.md` | Policy-set origin, local authority, adoption baselines, and upstream adoption and update links. | Task-specific reading triggers, detailed adoption procedures, and formal attribution notices. |
+| Agent policy | `docs/policy/agents.md` | Proportionate, idiomatic implementation; rules against inventing requirements, restrictions, guarantees, or policy. | Application contracts, architecture, testing policy, and contributor procedures. |
+| Documentation policy | `docs/policy/documentation.md` | Documentation ownership, audience boundaries, maintenance rules, and ADR/document lifecycle. | Application architecture or product behavior. |
+| Testing policy | `docs/policy/testing.md` | Test philosophy, risk sufficiency, boundaries, doubles, coverage, regressions, and test maintenance. | Subsystem behavior, application contracts, subsystem-specific test inventories, and implementation plans. |
+| Release policy | `docs/policy/release.md` | Versioning, compatibility, release integrity and immutability, release documentation, proportionate gates, publication controls, completion, and required versus optional checks. | Repository-specific commands, product installation guidance, CLI version semantics, historical release summaries, and CI implementation detail. |
+| Open source licensing and attribution policy | `docs/policy/oss.md` | License defaults and author choices, COPYRIGHT and attribution formats, license/notice links, discoverability, maintenance, and concrete exceptions. | Component entries, license texts, release commands, and dependency security review. |
+| Project copyright and license grant | `COPYRIGHT` | Project identity, copyright holders and years, SPDX expression, grant, warranty disclaimer, and third-party statement. | Complete license texts and upstream component attribution. |
+| Project license texts | `LICENSE` | Complete project and incorporated companion license texts, with required project-specific fields. | General licensing policy and upstream component attribution. |
+| Upstream component attribution | `docs/oss/attribution.md` | House Format component entries: identity, material scope and modifications where applicable, copyright, license, required notices, and license, notice, and source links. | General attribution policy, exception reasoning, research narratives, compliance commentary, and complete standard license texts. |
+| Standard dependency license texts | `docs/oss/licenses/` | Complete, verbatim standard license texts shared by applicable entries. | Project-specific upstream notices, attribution prose, and policy interpretation. |
+| Repository-specific release procedure | `docs/release.md` | Maintainer prerequisites, candidate preparation, validation and publication commands and automated guards, asynchronous deliverables, delivery status, optional consumer verification, and recovery. | General release requirements, product installation guidance, historical release summaries, and current compatibility contracts. |
+| Consumer guidance for the public Go API (if applicable) | `docs/consumers/` | Task-oriented public Go API usage, minimal examples, and consumer responsibilities. | HTTP wire semantics, external protocol contracts, internal implementation detail. |
+| Integration contracts for external components (if applicable) | `docs/integrations/` | External formats and protocols, upstream and downstream contracts, logical output paths and schemas, media types, and compatibility. | Physical runtime placement and lifecycle, internal transformations, CLI syntax, configuration defaults. |
+| Inventory of implemented internal packages | `docs/internal/overview.md` | Implemented packages and components, responsibilities, and links to focused internal docs. | Normative architecture, contributor reading policy, external contracts. |
+| Behavior and architecture of internal packages | Other files under `docs/internal/` | Implementation flow, collaborators, state transitions, package-local guarantees and failures, and relevant tests. | Global architecture invariants, configuration definitions and defaults, external schemas, operator procedures. |
+| Architectural decision records | `docs/adr/` | Significant decisions, context, alternatives, rationale, consequences, and supersession history. | Current behavior reference, implementation status, task sequencing. |
+| Future work and implementation status | `docs/roadmap/` | Proposed, accepted, deferred, or rejected work; implementation status, sequencing, and task breakdowns. | Implemented behavior reference and architectural decision rationale. |
+| Release-note history | `docs/releases/` | One checked-in summary per source release under the procedure; the note at its immutable tag is the release record. | Current commands, behavior, contracts, and compatibility definitions. |
+| Complete copyable artifacts | `examples/` | Maintained configuration, inputs, and other copyable or runnable files. | Field-by-field reference, command reference, prose explanation. |
 | Exact exported Go API | Go declarations and GoDoc, as APIs are implemented | Exported names, signatures, types, values, errors, and exact behavioral contracts. | Task-oriented consumer walkthroughs, implementation details, and future API proposals. |
-| Framework file formats | `docs/formats.md` | Prompt-definition and profile YAML fields, schema references, defaults, validation modes, built-in profiles, credentials, and file-to-request precedence. | Exported Go declarations, outbound wire behavior, internal parsing mechanics, and application configuration. |
-| Consumer guidance | `docs/consumers/`, when consumer workflows require dedicated guidance | Task-oriented use of implemented public APIs, minimal examples, and consumer responsibilities. | Exact exported declarations and internal mechanics. |
-| Durable integration contracts | `docs/integrations/`, when integrations exist | External formats and protocols, compatibility behavior, and upstream or downstream responsibilities. | Internal transformations and public Go declarations. |
-| Supplemental release guidance | None. `docs/releases/` may be used when a release benefits from a changelog or migration guide. | No canonical content. These files may briefly summarize release-specific changes, compatibility, and consumer migration paths, and may be corrected, consolidated, archived, or removed when no longer useful. | Public API and behavior contracts, formats, integrations, architecture, release procedure, and the authoritative annotated-tag release record. |
-| Implemented component inventory | `docs/internal/overview.md` | Current packages and components, their implemented responsibilities, and links to focused internal documents. | Normative architecture, contributor workflow, external contracts, and proposed components. |
-| Internal subsystem behavior | Other files under `docs/internal/`, when a subsystem needs durable detail | Implementation flow, internal collaborators and state transitions, package-local guarantees and failures, and relevant tests. | Global architecture invariants, public API definitions, and future package plans. |
-| Architectural decision history | `docs/adr/`, when repository-local decisions require records | Significant decisions, context, alternatives, rationale, consequences, and supersession history. | Current behavior reference, implementation status, and task sequencing. |
-| Temporary feature roadmaps | `docs/roadmap/`, while planned work needs coordination | Proposed or accepted scope, sequencing, gates, and implementation status. | Implemented behavior reference and durable decision rationale. |
-| Complete copyable artifacts | `examples/` | Valid inputs, Go programs, and other files intended to be copied or run. | Field-by-field reference, exact API declarations, and prose explanation. |
+| Framework file formats | `docs/formats.md` | Prompt-definition and profile YAML fields, schema references, defaults, validation modes, credentials, and file-to-request precedence. | Exported Go declarations, outbound wire behavior, internal parsing mechanics, and application configuration. |
 
-Conditional owners do not require placeholder files or directories. Create a
-consumer, integration, release, subsystem, ADR, roadmap, or example document
-only when the corresponding implemented interface, release, decision, planned
-effort, or maintained artifact exists.
+Create supporting documents for existing interfaces and responsibilities when
+substantive content is available.
 
 ## Boundary Rules
 
 ### Orientation
 
-The README owns project orientation. The development guide routes
-contributors. Architecture owns normative structure. The internal overview
-owns the current concrete component map. These documents may link to one
-another but must not maintain parallel package or behavior descriptions.
+Link among orientation, architecture, and internal documents; maintain each
+package or behavior description in its canonical owner.
 
 ### Public Contracts And Implementation
 
-Go declarations and GoDoc own exact exported API contracts once those APIs
-exist. Consumer and integration documents explain how to use those contracts
-for a task. Internal documents explain how Promptkit implements them. Internal
-documentation may identify a public type or external format as a dependency,
-but must link to its canonical definition rather than restate it.
+Go declarations and GoDoc own exact exported Go API contracts. The
+[framework format reference](../formats.md) owns prompt, profile, and schema
+formats; integration documents own external wire behavior. Consumer and
+internal documents link to those owners for exact definitions.
 
-The [framework format reference](../formats.md) owns exact prompt, profile, and
-schema-file contracts. Integration documents own external wire formats.
+### Releases
 
-### Supplemental Release Guidance
+The release policy defines requirements and invariants that every release must
+satisfy. The repository-specific release procedure defines the commands and
+workflow maintainers use to satisfy that policy. Release notes record history
+at immutable tags. Maintain current behavior in its canonical documentation,
+update it with the corresponding behavior changes, and keep release procedures
+consistent with adopted policy.
 
-Files under `docs/releases/` may provide changelog-style summaries and
-migration guidance for a particular release. They are navigation and
-orientation aids, not canonical owners of public APIs, behavior, formats,
-integrations, architecture, release procedure, or other durable facts. When a
-reader needs detail beyond a short release-specific note, the release document
-must link to the applicable canonical documentation rather than reproduce its
-contract.
+### Project Licensing And Upstream Attribution
 
-The annotated tag message required by the
-[release procedure](../release.md#write-the-release-note) remains the
-authoritative release record. Supplemental release documents may be corrected,
-consolidated, archived, or removed at any time when they are no longer useful,
-provided maintained documentation does not depend on them and the annotated
-tag record remains intact.
+README and release documents link to the canonical licensing and attribution
+materials under the [OSS policy](oss.md). The release procedure owns URL
+construction and publication mechanics.
+
+### Contracts And Implementation
+
+Integration and API documents define externally observable shapes and
+semantics. Internal documents explain how this application implements or consumes those
+contracts. Internal docs may name a field, file, or protocol to identify a
+dependency, but must link to its canonical contract for the definition.
 
 ### Security Topics
 
 This policy owns what documentation and examples may contain. Architecture owns
-library security boundaries and invariants. Public declarations and integration
-documents own consumer-visible security contracts. Internal documents own
-implementation mechanisms only.
+library security boundaries. Public declarations and integration documents own
+consumer-visible contracts; internal documents own implementation mechanisms.
 
 ## Architecture Decision Records
 
@@ -147,48 +148,47 @@ Use sequentially numbered ADR filenames such as
 6. alternatives considered;
 7. consequences.
 
-Use one of these statuses:
+Treat the decision content of an accepted ADR as immutable. When a decision
+changes, create a new ADR and update the earlier ADR's status to superseded.
+Rejected architectural alternatives belong in the ADR; rejected product ideas
+belong in the roadmap.
 
-- **Proposed:** the decision is under consideration and may change;
-- **Accepted:** the decision is approved, whether or not implementation is
-  complete;
-- **Rejected:** the proposed decision was considered and not adopted;
-- **Superseded:** a later accepted ADR replaces the accepted decision.
+## Maintenance
 
-A proposed ADR transitions to Accepted or Rejected. An Accepted ADR transitions
-to Superseded only when a later Accepted ADR replaces it. An ADR may be created
-as Accepted when the decision has already been made.
+When behavior changes, update its canonical owner in the same change. If
+ownership moves, remove the old definition and replace it with a link where
+navigation remains useful.
 
-Treat the decision content of an Accepted ADR as immutable. A changed decision
-requires a later ADR rather than a rewrite of the accepted record. A Superseded
-ADR must link to its replacement, and the replacement must link back. Rejected
-architectural alternatives belong in the ADR; rejected feature ideas belong in
-a roadmap when they need to be retained.
+Before completing documentation work:
+
+- verify affected behavior and examples;
+- check commands, flags, fields, defaults, schemas, and paths against their
+  implementation;
+- keep unimplemented behavior in the roadmap, subject to the ADR exception;
+- remove stale references and validate links;
+- confirm that non-owning documents summarize and link rather than redefine;
+- confirm that no secrets or sensitive private data were added.
+
+## Promptkit Documentation Scope
+
+Promptkit is a Go library, so exported GoDoc and framework formats replace
+application CLI, configuration, operations, and HTTP-service owners. Reading
+triggers belong in the development guide, not this topic map.
 
 ## Document Lifecycle
 
-Create durable current-state documentation with the implementation it
-describes. Update its canonical owner in the same change when behavior changes.
-If ownership moves, remove the old definition and leave a link where navigation
-remains useful.
-
 Roadmaps are temporary coordination documents. When their work is complete,
 record completion, move any still-useful decisions or contracts to their
 durable owners, update incoming links, and archive or remove the roadmap
 according to repository practice. Do not preserve completed roadmaps as a
 second current-state reference.
 
-Supplemental release documents may likewise be removed without preserving a
-replacement. Before removal, update maintained incoming links so current
-documentation does not depend on an optional historical guide.
+Retain release notes as historical records under the release policy. Do not
+apply temporary-roadmap retirement rules to published release notes.
 
-Before completing documentation work:
-
-- verify affected behavior and examples;
-- check commands, imports, declarations, defaults, schemas, and paths against
-  their implementation;
-- keep unimplemented behavior in a roadmap, subject to the ADR exception;
-- validate links and fenced examples;
-- confirm non-owning documents summarize and link rather than redefine;
-- remove stale or unsupported claims; and
-- confirm that no secrets or sensitive private data were added.
+For Promptkit ADRs, retain the existing Proposed, Accepted, Rejected, and
+Superseded statuses. Proposed records may become Accepted or Rejected;
+Accepted records may become Superseded, with links between the earlier record
+and its replacement. Records may be created as Accepted when the decision has
+already been made. These lifecycle rules preserve existing local document
+semantics.

docs/policy/meta.md

@@ -0,0 +1,27 @@
+# Policy Set Provenance
+
+**Base Policy Version:** 1.0
+
+**Last Updated**: 2026-09-27
+
+**Upstream Source:** [meta.md](https://gitea.maximumdirect.net/eric/policy/src/commit/d84a38ebab26f048cafbe59a2246700c35d29843/docs/policy/meta.md)
+
+## Origin And Local Authority
+
+This repository's policy set is adapted from
+[Standard Repository Policies](https://gitea.maximumdirect.net/eric/policy).
+The adopted local policies govern work in this repository. Upstream revisions
+become local policy through deliberate adoption, preserving local decisions.
+Each policy's `Upstream Source` identifies its adopted baseline.
+
+## Adoption And Updates
+
+For policy adoption, upstream revision reviews, and updates, consult the
+[upstream adoption and update guide](https://gitea.maximumdirect.net/eric/policy/src/branch/main/docs/adoption.md).
+That guide and the reference example collection are maintained upstream and
+accessed through links. Keep discovery links pointed at the current upstream
+project and guide; keep adopted-source and example links pinned to commits.
+
+The local [developer guide](../development.md) routes tasks to applicable
+policies. Formal credit and licensing notices for the copied policy material
+belong in the local [attribution document](../oss/attribution.md).

docs/policy/oss.md

@@ -0,0 +1,228 @@
+# Open Source Licensing And Attribution Policy
+
+**Base Policy Version:** 1.8
+
+**Last Updated**: 2026-11-02
+
+**Upstream Source:** [oss.md](https://gitea.maximumdirect.net/eric/policy/src/commit/d84a38ebab26f048cafbe59a2246700c35d29843/docs/policy/oss.md)
+
+## Purpose And Scope
+
+This policy defines project licensing conventions and a repeatable House Format
+for upstream attribution and license notices. It covers the project's own code
+and upstream components incorporated into or distributed with the application
+or library, including relevant transitive dependencies and copied code,
+documentation, or assets.
+Determine treatment from the component's terms and actual use.
+
+Consulting development guidance or using tooling alone does not require an
+attribution entry. Copying or adapting documentation for inclusion in the
+repository is different: preserve its applicable attribution and license terms.
+A material's role as development guidance does not exempt copied documents.
+Apply any express permissions for attribution-free reuse of example formatting.
+
+The baseline is a publicly available source repository with releases published
+alongside it. The format is independent of language, forge, and build system.
+It does not require attribution for every development tool, establish license
+compatibility, or resolve every obligation arising from use or distribution.
+
+## Canonical Locations
+
+The [documentation policy](documentation.md) assigns ownership:
+
+- `COPYRIGHT`: project name, copyright statements, SPDX expression, grant,
+  warranty disclaimer, and third-party statement.
+- `LICENSE` contains the complete, unmodified project license text, including
+  incorporated companion text. Dependency attribution does not relicense the project.
+- `docs/oss/attribution.md`: canonical component entries.
+- `docs/oss/licenses/`: complete, verbatim standard and applicable companion texts.
+
+Use recognizable license filenames, such as `Apache-2.0.txt` or `LGPL-3.0.txt`.
+A shared standard text may serve multiple entries. Keep explanations outside
+license texts and preserve distinct versions and exceptions.
+
+## Upstream Reference Examples
+
+Use the commit-pinned upstream examples to create actual project notices.
+Retain their links and keep the example collection upstream; keep applicable
+license texts locally as specified above.
+
+## Project Licensing
+
+Unless the author specifies otherwise, use **GPL version 3 or later**
+(`GPL-3.0-or-later`) for applications and **LGPL version 3 or later**
+(`LGPL-3.0-or-later`) for libraries. An author's explicit choice takes precedence:
+for example, a one-off script may use MIT. Such a choice requires no exception
+rationale, additional approval, or compensating controls under this policy.
+
+These defaults guide new license selection. Adopting this policy does not
+relicense existing code or override third-party terms. Preserve an existing
+license unless the author deliberately changes it. For mixed-purpose projects,
+follow the author's selected scope rather than inferring different licenses for
+individual packages.
+
+Use the upstream [application COPYRIGHT example](https://gitea.maximumdirect.net/eric/policy/src/commit/a469273adda336625f96b9d769ea27b33999f146/examples/licensing/application/COPYRIGHT)
+or [library COPYRIGHT example](https://gitea.maximumdirect.net/eric/policy/src/commit/a469273adda336625f96b9d769ea27b33999f146/examples/licensing/library/COPYRIGHT)
+to create the project’s root `COPYRIGHT` file.
+Use verified project names, contribution years, and copyright holders.
+For another selected license, adapt the grant, SPDX expression, disclaimer, and
+text references together to reflect the author's selected grant.
+
+For GPLv3 projects, copy the complete GPLv3 text into `LICENSE`. For LGPLv3
+projects, place the complete LGPLv3 text followed by the complete GPLv3 text in
+`LICENSE`; the LGPL incorporates the GPL terms. Keep the texts unmodified and
+place project-specific statements in `COPYRIGHT`. For the default grants, the
+grant and SPDX expression in `COPYRIGHT` must expressly identify `or-later`;
+follow the author’s choice if a different grant is selected.
+
+For other licenses, provide their complete text and any required project-specific
+fields; the standard reference collection remains unchanged. Keep `COPYRIGHT`,
+`LICENSE`, SPDX expressions, and README descriptions consistent with the actual
+grant. This convention introduces no blanket requirement for per-file headers;
+preserve notices or add them where the applicable terms require them.
+
+Identify separately licensed documentation or assets in the project grant, with
+links to their attribution and license information. An application's default
+code license does not override the terms of incorporated documentation.
+
+## Upstream Attribution House Format
+
+Begin `docs/oss/attribution.md` with `# Open Source Attribution Notices`,
+followed directly by component entries. These comprise the complete document.
+
+Each entry presents these five fields in order, with blank lines between them:
+
+1. **Project name** as a level-two heading. An optional identification paragraph
+   may follow, such as `Selected module: MODULE at VERSION.` with the module and
+   version in backticks. For copied documentation, a concise `Material:` sentence
+   may identify the affected files and whether they were adapted or copied
+   without modification. A factual qualifier such as `(transitive dependency)`
+   is permitted. Limit optional identification prose to factual component and
+   material scope.
+2. **Upstream copyright statement or statements**, preserving their wording.
+   Put multiple statements in separate paragraphs. Include this field when the
+   selected upstream source supplies an applicable statement; otherwise, omit it.
+3. **SPDX license identifier or expression**, when applicable, using
+   `SPDX-License-Identifier: EXPRESSION`.
+4. **License and notice links**, beginning `License and notices:`, separated
+   by ` | `, and ending with a period. Label local standard texts by license
+   name, such as `MIT` or `Apache 2.0`. Label upstream files by their actual
+   filenames, such as `COPYRIGHT`, `LICENSE`, `NOTICE`, or `PATENTS`.
+5. **Source repository link**, rendered exactly as:
+
+   ```markdown
+   [Source](COMMIT_URL).
+   ```
+
+Use the upstream [attribution example](https://gitea.maximumdirect.net/eric/policy/src/commit/a469273adda336625f96b9d769ea27b33999f146/examples/licensing/attribution.md)
+to create the project’s `docs/oss/attribution.md`.
+The example illustrates omitted fields and multiple copyrights and licenses.
+Replace the uppercase placeholders and illustrative URLs with actual component
+information. Local license links are relative to the destination
+`docs/oss/attribution.md`; copy the applicable texts into its `licenses/`
+directory. Do not copy example entries for components the project does not use.
+
+These are five fields, not a line limit. Omit the SPDX field when no applicable
+identifier exists; identify nonstandard terms through the linked text. Preserve
+licensing alternatives and exceptions, and record a selected option when
+relevant through the factual license identification. Do not invent an identifier
+or silently resolve an unsettled licensing choice.
+
+Keep entries factual. Do not add agent-authored explanations of license
+expressions, file-level license scope, research performed, retained source
+notices, or compliance conclusions. Avoid filler such as "none is invented
+here" and embellished link labels such as "upstream LICENSE", "additional
+patent grant", or "at the imported commit". Preserve actual required upstream
+content; this rule excludes editorial additions, not required notices. Policy
+interpretation and exception reasoning belong in the adopted OSS policy or an
+appropriate decision record, not the attribution document.
+
+## License Texts And Upstream Links
+
+Link standard license texts to their local copies. Link project-specific
+notices to their upstream copies. Every upstream repository or notice link must
+identify the exact commit corresponding to the
+imported version, including the selected source revision for copied material.
+
+Classify material according to its contents. An upstream `LICENSE`,
+`COPYING`, or similar file may contain project-specific notices or exceptions
+in addition to standard wording. Link both the local standard text and the
+commit-specific upstream file when needed. Make companion license texts directly
+reachable from the entry.
+
+For project-specific license and attribution material, this policy accepts
+direct links to the applicable upstream texts as satisfying reproduction,
+inclusion, or copy requirements. This is the adopted house interpretation.
+Agents must apply it without adding local-copy requirements or seeking repeated
+approval solely because a license uses those terms.
+
+This interpretation concerns notice reproduction. Concrete obligations such as
+marking modified files or supplying corresponding source receive their own
+applicable treatment. Preserve required notices already present in incorporated
+source files; the central attribution entry does not authorize stripping them.
+
+## Discoverability
+
+End the project's README with two friendly statements linking to its actual
+grant in `COPYRIGHT` and the attribution document. For example, for a project
+with the stated license:
+
+```markdown
+PROJECT_NAME is licensed under the [LGPL v3.0 or later](COPYRIGHT).
+
+[Open Source attribution notices](docs/oss/attribution.md) are available.
+```
+
+Each new checked-in release document and its hosted download page must end
+with links whose visible text is `COPYRIGHT | LICENSE | ATTRIBUTION`. Use absolute
+URLs to those files at the release's immutable tag or corresponding immutable source
+revision so the links work in both locations. This release navigation is distinct
+from upstream component links, which must use the imported commit.
+
+Reuse the checked-in release document as the hosted release body where supported.
+The release procedure owns concrete URLs and publication mechanics for delivering
+the links through the selected channel. Do not maintain duplicate notice lists.
+These conventions do not by themselves require separate downloadable license
+assets, source bundles, or a different binary package format.
+
+## Maintenance And Exceptions
+
+Update project licensing statements together when the author changes the grant
+or its scope. Update upstream notices in the change that adds, removes, upgrades,
+copies, or changes the distribution of a relevant component. Check the selected source's license
+and notices rather than assuming they are unchanged. Update affected entries,
+commit links, and texts; retain historical versions through Git.
+
+Check affected statements, faithful standard-text copies, and link destinations
+as part of ordinary change review. For a future release tag, check intended file
+paths against the candidate source; the tag need not resolve remotely before
+publication. Do not add publication-time network checks, CI polling, recurring
+legal sign-offs, mandatory scanners, or a full software bill of materials merely
+to implement this policy.
+
+Use the House Format routinely. Raise an exception only for a concrete obligation
+or unresolved rights question it does not address, including any arising from a
+particular distribution channel. Record the issue and selected treatment in the
+adopted OSS policy or an appropriate decision record. Attribution entries
+contain the resulting factual identification and required material, without
+repeating that reasoning. Resolve the issue before the affected use or
+distribution without reopening settled choices.
+Common ownership of projects is neither required nor a blanket exception.
+
+Apply the formatting convention prospectively. Do not rewrite published tags,
+notes, binaries, or checksums to retrofit it. Address actual omissions deliberately
+under the applicable terms and the [release correction policy](release.md#immutability-and-failure-recovery).
+Ordinary attribution maintenance and release links are preparation work, not new
+release approval gates.
+
+## Promptkit Adoption
+
+Promptkit retains its LGPL-3.0-or-later library grant in
+[COPYRIGHT](../../COPYRIGHT). Adapted policy documentation has the separate
+scope identified there and in [attribution](../oss/attribution.md).
+
+Promptkit publishes source-only Go module tags and Gitea release pages, with no
+binaries. Each new hosted page reuses the complete checked-in release note,
+including its licensing footer. The [release procedure](../release.md) owns
+source publication, hosted delivery, and concrete footer URLs. Previously
+published notes and tags retain their original records.

docs/policy/release.md

@@ -0,0 +1,292 @@
+# Release Policy
+
+**Base Policy Version:** 1.6
+
+**Last Updated**: 2026-11-02
+
+**Upstream Source:** [release.md](https://gitea.maximumdirect.net/eric/policy/src/commit/e833faddf120c789a9b45348edaf75720de16348/docs/policy/release.md)
+
+## Purpose
+
+This policy defines a proportionate workflow for validating and publishing an
+intended source revision under a fresh version. Releases should require little
+routine intervention, avoid repeated work, and preserve published contents.
+
+Adopt stronger verification or reproducibility requirements through an explicit
+project-specific decision justified by the intended audience and support needs.
+
+The repository-specific commands belong in `docs/release.md`. Historical release
+summaries belong in `docs/releases/`. The procedure must agree with the adopted
+policy and link to it rather than introduce additional release requirements.
+
+## Release Model And Completion
+
+A source release identifies one exact committed source revision through an
+immutable version tag and a version-matched, checked-in release note. The
+primary release branch is `main` unless the adopted policy specifies otherwise.
+The procedure must name the authoritative publication remote.
+
+A successful push of the intended tag, with its source commit available on the
+authoritative primary branch, completes the release. The source commit may be
+pushed before the tag or published together with it through an atomic push.
+The procedure must choose and document the supported approach.
+
+Release commands must finish at successful publication. They must not wait for
+or poll CI, require successful CI status, inspect downloadable artifacts, or
+check consumer installation or Go module availability as completion gates.
+A failed or ambiguous push requires resolving the authoritative remote state
+before declaring success or retrying; successful pushes do not require a
+separate remote read-back gate.
+
+Repositories may publish source only. Additional binaries, images, packages,
+checksums, or other declared deliverables may be published asynchronously from
+the tagged source. The procedure must identify those deliverables and the
+responsibility of their automation. Published deliverables must be traceable
+to that source and identify the same version.
+
+Track source-release completion and artifact delivery separately. Report delivery
+failures and optional investigations under artifact-delivery status.
+
+## Versioning And Compatibility
+
+Stable releases use Semantic Versioning in the form `vMAJOR.MINOR.PATCH`, with
+no leading zeroes. Prereleases require an explicit adopted policy defining
+their format and use.
+
+The selected version must be unused locally and on the authoritative remote.
+Do not require queries to every configured remote or publication service.
+A version known to have already been published must not be reused, even if the
+normal guards do not inspect that channel. Adding a distribution channel
+requires a deliberate update to the policy and procedure where needed.
+
+For releases at or after `v1.0.0`:
+
+- a major release may make intentional incompatible changes;
+- a minor release may add backward-compatible behavior or raise the Go consumer
+  minimum as described below; and
+- a patch release may make backward-compatible fixes.
+
+Before `v1.0.0`, a minor release may change a documented public, operational,
+configuration, data, integration, or compatibility contract when the release
+note explains the impact and required consumer action. A patch release must
+not intentionally break those contracts within its minor line.
+
+Additional compatibility promises and deliberate exceptions belong in the
+adopted policy or canonical contract documentation. Release notes explain
+changes; they do not redefine current compatibility contracts.
+
+## Consumer Minimum And Release Toolchain
+
+Distinguish the minimum toolchain supported for consumers from the toolchain
+maintainers use to build and validate a release. Record the consumer minimum in
+the language's standard metadata and explain its support meaning in canonical
+documentation. Define the selected release toolchain and its setup commands in
+`docs/release.md`. Toolchain selection and upgrade cadence are repository-specific
+choices.
+
+For Go repositories, normally declare the initial stable release of the selected
+series as the consumer minimum, such as `go 1.26.0`. Use a higher patch minimum
+when a dependency or an identified correctness fix requires it, and document
+the reason.
+
+For release-policy purposes, successful build and release validation on the
+selected Go patch release are sufficient to presume support across that Go
+series, subject to the declared consumer minimum and concrete evidence of
+incompatibility. Investigate specific compatibility concerns as they arise.
+
+Use the selected release toolchain for candidate validation and release builds,
+including CI jobs performing those tasks. A newer validation patch version alone
+leaves the consumer minimum unchanged.
+
+A minor release may raise the Go consumer minimum, including from one Go series
+to the next, both before and after `v1.0.0`. This is an explicit exception to the
+backward-compatibility requirement for minor releases. Patch releases preserve
+the consumer minimum. Explain an increase's impact and required consumer action
+in the release note. Other incompatible changes remain subject to the versioning
+rules above.
+
+## Release Documentation And Preparation
+
+Every new release must include `docs/releases/<version>.md` in the tagged
+source tree. Use the complete version tag as the filename and begin with a
+title containing the product name and that exact version.
+
+Include a short summary and any relevant compatibility impact or upgrade
+instructions. Choose sections suited to the release and keep the note
+proportionate to the change.
+
+Keep exact current contracts in their canonical documents and update affected
+documentation with the behavior change. Link to those documents from the note
+where useful. If annotated tags are used, derive their summary from the
+checked-in note.
+
+Include the release footer defined by the [OSS policy](oss.md#discoverability)
+in each new release document and preserve it on the hosted download page. The
+release procedure defines the concrete URLs and publication mechanics. Prepare
+these links with the release note; they do not add a network check or approval
+gate.
+
+Handle version selection, release-note content, and retirement of completed
+roadmaps during preparation. Preserve planned and deferred work and repair
+affected links when retiring roadmaps. Reuse ordinary change review for
+unintended files, credentials, and private data. Scope release-content review
+to the tagged source.
+
+## Candidate Validation
+
+Provide one publication command that validates the final committed candidate,
+checks publication state, and publishes the intended refs. A validation-only
+mode or command may be provided for diagnosis, but must not be a required step
+before publication repeats the same checks.
+
+Run the complete deterministic, offline default test suite once for the final
+candidate, following the [testing policy](testing.md). For repositories that
+publish builds, verify the required build targets. Select those targets for
+the application's actual support commitments; native testing on each platform
+and access to particular hardware are not default release prerequisites.
+
+Additional gates must address a concrete risk that the existing checks do not
+adequately cover. Do not automatically require linting, race testing, module
+hygiene audits, documentation audits, credential scans, example validation, or
+expanded platform matrices solely because they are possible. Retain such
+checks when their distinct protection justifies their cost. A one-time migration
+check must not silently become a permanent release requirement.
+
+Release tests must not require real credentials, mutable application services,
+paid APIs, provider probes, or inference calls. The procedure may describe dependency
+preparation separately from offline validation.
+
+Avoid duplicate validation across release phases. Successful checks for the
+same final candidate need not be repeated solely to satisfy another phase;
+any reuse must clearly apply to that exact candidate. Persistent validation
+caches are not required. If the candidate changes, validate the new candidate.
+Cheap publication-state checks may be repeated immediately before pushing
+because local or remote state may have changed during validation.
+
+Automate mechanical guards. Human review gates are strongly disfavored: add one
+only through an explicit project-specific policy decision explaining the
+concrete risk and why ordinary change review or automation is insufficient.
+Do not add routine confirmations for checks the command can perform itself.
+
+## Publication Guards
+
+Automate the checks needed to publish the intended source under a fresh version:
+
+- the version has the adopted format and is unused locally and on the
+  authoritative remote;
+- the worktree and index are clean on the primary release branch;
+- the source being published is the exact candidate that passed validation,
+  including its matching release note;
+- the remote primary branch already identifies that candidate, or an atomic
+  source/tag push will advance it to that candidate without discarding remote
+  history; and
+- the tag identifies the validated candidate and only the intended refs are
+  published.
+
+Fetch the authoritative branch and tag state as needed for these checks.
+Do not force branch or tag updates to bypass a failed guard. Publish explicit
+refs rather than all local tags. Do not require CI success, publication-service
+availability checks, or a separate manual confirmation for each guard.
+
+## Immutability And Failure Recovery
+
+Published version tags and published contents are immutable. Do not move,
+overwrite, recreate, or delete a published tag, or replace a published artifact,
+checksum, signature, provenance statement, or release note under that version.
+The checked-in note at the immutable tag is that release's historical record.
+
+If validation fails before publication, correct and commit the candidate and
+rerun the publication command. If a push fails or its outcome is ambiguous,
+inspect the authoritative remote before retrying. Remove an unpublished local
+tag only after confirming it was never published or used by a release channel.
+An unavailable remote leaves publication status unknown; do not overwrite a tag
+to resolve uncertainty.
+
+Transient network, authentication, CI, or consumer-availability failures do not
+by themselves require a new version. Asynchronous delivery may be retried from
+the same immutable tag. Missing artifacts may be published later, provided
+existing published contents are not replaced and the additions remain consistent
+with the tagged source and any already-published checksum manifest. If that
+cannot be achieved, publish a new version.
+
+A defect requiring changes to published source or contents requires a corrected
+commit, a new semantically appropriate version, a matching release note, and
+the release procedure again. Leave the original release intact. A channel's
+yank, deprecation, or advisory mechanism may warn consumers away from a defective
+release without replacing its contents or reusing its version.
+
+## Repository Release Procedure
+
+Maintain `docs/release.md` as a short, executable maintainer runbook. It must
+identify:
+
+1. prerequisites, selected release toolchain, primary branch, and authoritative
+   remote;
+2. version and release-note preparation;
+3. one publication command with its selected local checks and automated guards;
+4. declared deliverables, required build targets, and automation responsibilities;
+5. the successful-push completion point and separately reported delivery status;
+   and
+6. failure handling, ambiguous-push recovery, and asynchronous delivery retries.
+
+Use copyable commands where practical. Keep optional diagnostics and consumer
+verification clearly optional. The procedure must not turn preparation work,
+ordinary change review, or post-publication investigation into additional
+release gates.
+
+## Promptkit Release Configuration
+
+The release-note requirements were adopted on 2026-09-23 and licensing
+navigation on 2026-09-26. Apply them prospectively; previously published notes
+and tags retain their original records.
+
+Promptkit publishes source-only Go modules from main to origin on Gitea, using
+annotated stable SemVer tags. The source commit is pushed through main before
+the tag. After source publication, maintainers use ReleaseKit locally to deliver
+a Gitea release page containing the complete checked-in note, including its
+licensing footer. There are no binaries or other deliverables. Source publication
+and hosted delivery have separate outcomes; delivery credentials and service
+availability do not gate source-release completion. The
+[release procedure](../release.md) owns commands and recovery.
+
+Release preparation uses the latest Go installation available in the development
+environment to build the maintainer release executable and validate the
+candidate, including the consumer examples. Keep that same local toolchain
+active throughout those operations.
+
+For a minor release, advance Promptkit's consumer minimum to the initial stable
+release of that Go series, retaining a higher patch minimum when required by a
+dependency or an identified correctness fix. Patch releases preserve the
+existing consumer minimum. Commit any module update before validating the final
+candidate. The [release procedure](../release.md#prepare-the-candidate) owns
+setup commands, and the consumer guide explains
+[Go compatibility](../consumers/pkg-promptkit.md#go-compatibility).
+
+Use a title containing `Promptkit` and the exact version, followed by a short
+summary and any relevant compatibility impact or upgrade instructions. The
+note is committed with the candidate and supplies the annotated tag message.
+No fixed set of section headings is required.
+
+The [publication command](../../scripts/release.sh) delegates to the released
+ReleaseKit tool selected in `go.mod`. The committed
+[profile](../../releasekit.json) selects the race-enabled Go suite, full vet,
+committed Go formatting, local Markdown links, workspace and filesystem
+replacement inspection. The Go suite includes the offline semantic checks for
+the three maintained consumer examples. It also rejects the configured local
+delivery token path if committed; this narrowly scoped check is not a general credential
+scan. These are the selected checks for the final candidate; a separate
+ordinary test run or standalone build is not an
+additional release gate. The upstream tool owns candidate, remote, and tag
+publication checks. It returns native status 0 for confirmed publication, 1
+for a known validation or publication failure, 2 for invalid input or profile,
+and 3 when the push outcome is unknown. For status 3, resolve the exact
+captured publication endpoint before retrying; the repository-specific
+[procedure](../release.md#failure-recovery) links to the upstream recovery
+instructions. Do not wait for CI or Go module availability. Licensing footer
+URLs are preparation work, not a network gate.
+
+Hosted delivery is a separate post-publication step for new releases. It reuses
+the immutable tag and note without rerunning candidate validation or waiting
+for CI. Retry failed delivery against that same version after resolving any
+uncertain hosted write; preserve published tags and hosted content. Report
+delivery status separately from the completed source release.

docs/policy/testing.md

@@ -1,39 +1,32 @@
 # Testing Policy
 
+**Base Policy Version:** 1.3
+
+**Last Updated**: 2026-09-27
+
+**Upstream Source:** [testing.md](https://gitea.maximumdirect.net/eric/policy/src/commit/d84a38ebab26f048cafbe59a2246700c35d29843/docs/policy/testing.md)
+
 ## Purpose
 
 Our tests exist to make **incorrect changes expensive and correct changes cheap**.
 
-We do not optimize for test count, line coverage, exhaustive isolation, or the fewest possible tests. We optimize for sufficient confidence in important behavior while imposing as little unnecessary friction as possible on future development.
+We optimize for sufficient confidence in important behavior while minimizing friction on future development.
 
 ## Every test has a cost
 
-Testing is not an unqualified good. Every test imposes both an immediate cost and a continuing lifetime cost.
-
-A test must be:
-
-- written and reviewed;
-- understood by future maintainers and coding agents;
-- executed in maintainer-run validation;
-- diagnosed when it fails;
-- updated when legitimate behavior changes;
-- maintained as fixtures, APIs, and dependencies evolve; and
-- removed or rewritten when it becomes redundant, brittle, misleading, or obsolete.
-
-Tests also create cognitive and architectural friction. They can constrain refactoring, duplicate policy, slow feedback loops, add noise to failures, and cause harmless implementation changes to require unrelated edits across the suite.
+Every test has an immediate and continuing cost: writing, review, execution,
+diagnosis, fixture maintenance, and updates when legitimate behavior changes.
+Tests can also constrain refactoring, duplicate policy, and add failure noise.
 
 A test is warranted only when the confidence it provides justifies these costs.
-
-Apply this cost-benefit analysis at two levels:
+Evaluate that tradeoff at two levels:
 
 1. **Per test:** What realistic defect does this test detect, how consequential would that defect be, and is that protection worth the test's lifetime cost?
 2. **Across the suite:** Does this collection provide materially more confidence than a smaller, simpler suite would?
 
-The preferred test suite is a **lean suite that provides sufficient confidence in the risks that matter, without redundant or low-value tests**. We seek sufficient confidence with the least unnecessary testing friction, not the fewest possible tests.
+Some friction is intentional. Tests should make dangerous changes—such as breaking compatibility, corrupting data, violating security boundaries, or reintroducing subtle bugs—require deliberate review. Keep ordinary internal changes inexpensive.
 
-Some friction is intentional. Tests should make dangerous changes—such as breaking compatibility, corrupting data, violating security boundaries, or reintroducing subtle bugs—require deliberate review. They should not make ordinary internal changes needlessly expensive.
-
-The cost of a test is not a reason to omit testing by default. Do not cite maintenance cost abstractly. When omitting a plausible test, be able to state why the protected failure is low-risk, already covered, obvious, reversible, or cheaper to detect elsewhere. For consequential, subtle, or difficult-to-observe behavior, the presumption should favor testing.
+Evaluate maintenance cost against the specific failure being protected. When omitting a plausible test, be able to state why the protected failure is low-risk, already covered, obvious, reversible, or cheaper to detect elsewhere. For consequential, subtle, or difficult-to-observe behavior, the presumption should favor testing.
 
 ## Default testing style
 
@@ -47,21 +40,41 @@ Use a **classical/Detroit-style** approach:
 
 Examples of appropriate seams include clocks, randomness, subprocesses, remote APIs, object storage, email, and paid LLM calls.
 
+## Promptkit Test Boundaries
+
+Promptkit is a library without a CLI or hosted CI. The execution requirements,
+facade test ownership, public error guidance, and offline consumer examples
+below tailor the standard to that boundary.
+
 ## Test execution requirements
 
 Promptkit currently uses maintainer-run validation rather than hosted CI.
-Maintainers run the complete local workflow in the
-[development guide](../development.md#maintainer-validation) before accepting
-changes. That guide is the canonical owner of exact commands, formatting,
-documentation-link validation, and repository-hygiene checks.
+Ordinary work requires validation proportionate to its scope and risk. Do not
+repeat successful checks for unchanged content merely to create a commit.
+Release candidates require the complete local workflow in the
+[development guide](../development.md#maintainer-validation). That guide routes
+contributors to the executable checks for formatting, documentation links,
+and repository hygiene.
 Introducing hosted CI later would supplement, not silently redefine, this
 documented validation model.
 
-Maintainer validation must include ordinary and race-enabled package tests,
-static analysis, a complete build, and execution of both maintained offline
-consumer examples. The preparation example protects assembled preparation and
+The selected ReleaseKit profile runs the race-enabled package suite and full
+vet, checks committed Go formatting, local Markdown links, and module
+distribution hygiene. The Go suite includes semantic assertions for the three
+maintained offline consumer examples. ReleaseKit owns this shared validation
+machinery and publication workflow. The test run builds the packages and
+exercises the examples. Separate ordinary tests and a
+standalone build are not additional release gates. Ordinary tests remain useful
+during development. The preparation example protects assembled preparation and
 inspection behavior. The execution example separately protects assembled
-`Run`, injected-client, validation, usage, and result behavior.
+`Run`, injected-client, validation, usage, and result behavior. The observation
+example protects assembled built-in retries, structured proposal and assessment
+runs, and application-owned interpretation of metadata.
+
+The standalone `python3 scripts/test-release.py` command exercises Promptkit's
+ReleaseKit wiring against disposable local repositories. It is focused
+development validation when release tooling changes, not part of the default
+Go test suite or a recurring release check.
 
 Tests in the default suite must be deterministic, offline, and independent of
 real credentials. They must not invoke paid APIs, use live network
@@ -87,7 +100,7 @@ Use each test type where it protects a distinct risk:
   interaction, while replacing live or nondeterministic external boundaries.
 - External-package root tests exercise the public facade as a Go consumer,
   while internal package tests own focused implementation behavior.
-- The maintained offline preparation and execution examples protect distinct
+- The maintained offline preparation, execution, and observation examples protect distinct
   representative assembled consumer workflows without contacting a model
   provider.
 - Fixtures should be minimal, synthetic, versioned with the behavior they
@@ -99,7 +112,7 @@ Use each test type where it protects a distinct risk:
 
 ## What deserves tests
 
-Prioritize tests for:
+To the extent the following features are present in the application, prioritize tests for:
 
 1. Public and package-level contracts.
 2. Domain rules and important invariants.
@@ -107,17 +120,29 @@ Prioritize tests for:
 4. Failure handling, cancellation, retries, recovery, and partial success.
 5. Serialization, schemas, compatibility, and round trips.
 6. Previously observed or plausible regressions.
-7. Representative integration and consumer workflows.
+7. Representative integration and end-to-end workflows.
 
-A package-level contract is behavior relied upon by another package or major collaborator, not every observable detail of a package implementation.
+A package-level contract comprises behavior relied upon by another package or major collaborator.
 
-For behavior involving **data integrity, destructive operations, compatibility, security, concurrency, idempotency, or recovery**, presume that durable tests are required unless the behavior is already credibly protected at another layer.
+Choose tests for requested behavior and established contracts. Tests must not
+introduce new restrictions, guarantees, or threat scenarios merely because they
+could be asserted. Follow the [agent policy](agents.md) when deciding whether a
+behavior belongs to the contract; current implementation behavior alone is not
+sufficient authority.
+
+For established behavior involving **data integrity, destructive operations,
+compatibility, security, concurrency, idempotency, or recovery**, favor durable
+tests for realistic failures whose consequences warrant protection. Identify
+the failure and the existing contract it would violate. These categories call
+for attention, not an automatic test inventory or permission to strengthen the
+application's guarantees. Account for protection already provided at other layers.
 
 Do not add tests merely because a function, branch, or line exists. Do not add a test when the same meaningful risk is already adequately protected elsewhere.
 
 ## Choose the right test boundary
 
-Test through the narrowest stable boundary that expresses the behavior clearly.
+Choose the narrowest stable boundary that expresses the behavior clearly and
+provides durable confidence with minimal incidental coupling.
 
 This is often the package API, but it may instead be:
 
@@ -125,30 +150,15 @@ This is often the package API, but it may instead be:
 - a package-level operation when several internal collaborators jointly produce the behavior; or
 - a larger integration boundary when correctness emerges from interaction with a real dependency.
 
-Do not force all behavior through oversized consumer-workflow tests. Do not
-test every private helper merely because it exists. Choose the boundary that
-gives durable confidence with the least incidental coupling.
-
 ## Test behavior, not implementation
 
-A test should protect a decision, contract, or invariant—not memorialize the current implementation.
+A test should protect a selected decision, contract, or invariant.
 
-Before adding or retaining a test, ask:
+A test is suspect when it primarily detects changes to private constants or
+helpers, equivalent operation ordering, incidental formatting, equivalent
+algorithms, or internal structure that preserves behavior.
 
-> What realistic defect would this test catch?
-
-A test is suspect when its main purpose is to detect that someone:
-
-- changed an internal constant;
-- renamed or split a private helper;
-- reordered equivalent internal operations;
-- changed incidental formatting;
-- replaced one correct algorithm with another; or
-- refactored internal object structure without changing behavior.
-
-Refactoring should normally require no test edits unless the refactored structure is itself part of the contract.
-
-A test can be factually correct and still have negative value. Accurately describing current behavior is not enough; the protected behavior must be important enough to justify the future friction.
+Behavioral tests should normally remain stable through internal refactoring that preserves the contract.
 
 ## Expected effects of different changes
 
@@ -165,13 +175,13 @@ A test failing is not the same as a test needing to be edited. Many tests may co
 
 ## Separate mechanism from policy
 
-Configurable thresholds and defaults must not be duplicated throughout the test suite.
+Derive expectations from test-controlled configuration and relationships.
 
-The following fragments are illustrative rather than standalone Go programs.
-Do not encode an internal concurrency limit indirectly:
+For this example, assume the selected contract rejects a request to start more
+processes than a configured limit. Do not encode its production default indirectly:
 
 ```go
-// Production policy:
+// Production default:
 const maxConcurrency = 4
 
 // Brittle test:
@@ -189,37 +199,40 @@ require.NoError(t, runner.Start(limit))
 require.ErrorIs(t, runner.Start(limit+1), ErrTooMuchConcurrency)
 ```
 
-The test should prove:
+For this selected contract, the test should prove:
 
 - the configured limit is accepted; and
 - one beyond the configured limit is rejected.
 
+Rejection is not a general requirement for concurrency control. If the selected
+contract queues excess work or limits active workers while accepting larger
+batches, test that behavior instead. Do not add rejection or a configurable
+limit solely to reproduce this example.
+
 The production default should be tested exactly only when its literal value is itself a public, operational, safety, protocol, or compatibility requirement.
 
-Apply the same rule to limits, timeouts, capacities, retry counts, and ranges: test relationships and behavior, not duplicated literals.
+Apply the same rule to limits, timeouts, capacities, retry counts, and ranges: test relationships and behavior.
 
-For concurrency limits, test both kinds of behavior when relevant:
+For established concurrency limits, distinguish these behaviors where the contract requires them:
 
-1. **Configuration enforcement:** invalid or excessive requested values are handled correctly.
+1. **Configuration handling:** values are accepted or rejected according to the selected contract.
 2. **Runtime enforcement:** observed peak concurrency never exceeds the configured limit.
 
-Use a test-controlled limit and measure the behavior relative to that limit. Do not merely assert today's default value.
+Use a test-controlled limit and measure the behavior relative to that limit.
 
-## Avoid semantic duplication across layers
+## Assign each behavior a test owner
 
-Each behavior should have a clear test owner.
+Where these layers exist, assign ownership as follows:
 
 - Parser tests own parsing cases.
 - Validator tests own validation rules.
 - Domain tests own transformations and invariants.
-- Boundary tests own external integration behavior.
+- Adapter tests own external integration behavior.
 - Orchestrator tests own coordination and failure propagation.
-- Consumer-workflow tests prove that representative assembled library use
-  works.
+- External-package root tests own the public Go facade contract.
+- Consumer-workflow tests prove that representative assembled library use works.
 
-Higher-level tests should not repeat every lower-level case. A single intentional policy change should not require unrelated edits across many test files.
-
-Tests that are individually reasonable may still be collectively redundant. Evaluate the marginal value of each additional test in light of the protection already provided by the rest of the suite.
+Give each layer its distinct responsibility and keep expectation updates localized to the contract being changed.
 
 ## Use test doubles deliberately
 
@@ -240,111 +253,78 @@ Mocks are appropriate when the contract includes facts such as:
 - an expensive API is called no more than once; or
 - a security audit event is emitted.
 
-Do not use mocks merely to isolate every object or reproduce the implementation's call graph.
-
 ## Go-specific guidance
 
-Use:
+Select techniques by behavior and risk:
 
 - table-driven tests for meaningful behavioral categories and boundaries;
 - `t.TempDir()` for real filesystem behavior;
 - `httptest.Server` for realistic HTTP interactions;
-- fuzz tests for parsers, normalization, path handling, and broad input spaces;
+- fuzz tests when exploring broad input spaces, such as parsers or path handling, adds useful protection beyond representative cases;
 - golden files only when the complete output is intentionally stable;
 - integration tests where correctness depends on component interaction; and
-- a small number of representative consumer-workflow tests.
+- a small number of representative end-to-end tests when assembled workflows expose meaningful risks not adequately covered at narrower boundaries.
 
-Avoid exact error-string assertions unless the wording is itself contractual. Prefer `errors.Is`, `errors.As`, typed errors, or structured error fields.
+Prefer `errors.Is`, `errors.As`, typed errors, or structured error fields. Use exact error-wording assertions only when the wording is contractual.
 
 At public API boundaries, prefer stable error identity, structured values, and
-the smallest semantic fragment needed to identify the failure. Do not snapshot
-complete diagnostic wording unless it is contractual.
+the smallest semantic fragment needed to identify the failure. Use complete
+diagnostic snapshots only when the wording is contractual.
 
-Golden-file updates must require an explicit local flag. Ordinary validation
-runs must never update golden files automatically, and maintainers must inspect
-the semantic diff before accepting an update.
+Golden-file updates must require an explicit local flag. Ordinary validation runs must never update golden files automatically, and reviewers must inspect the semantic diff before accepting an update.
 
-Keep tests readable and direct. Test helpers and fixture frameworks must earn their own maintenance cost; do not build elaborate test infrastructure for small or isolated needs.
+Keep tests readable and direct. Size test helpers and fixture frameworks to the maintenance benefit they provide.
 
 ## Coverage
 
-Coverage is a diagnostic, not a target.
+Use coverage to locate untested critical branches and unexpectedly weak packages. Evaluate test quality through meaningful behavioral protection and choose coverage according to risk.
 
-Use it to find untested critical branches and unexpectedly weak packages. Do not write low-value tests solely to increase a percentage, and do not infer test quality from coverage alone.
-
-Pure domain logic will often warrant higher coverage than facade wiring or
-external adapters. Uneven coverage is acceptable when it reflects risk.
-
-Increasing coverage is valuable only when the newly covered behavior protects a meaningful risk at an acceptable cost.
+Pure domain logic will often warrant higher coverage than facade wiring or external adapters. Uneven coverage is acceptable when it reflects risk.
 
 ## Regression tests
 
 A bug fix should normally include a regression test that fails before the fix and passes afterward.
 
-Retain the test when the defect could realistically recur and its consequences justify the ongoing cost. Prefer the narrowest durable test of the violated contract or invariant; do not preserve accidental implementation details from the original bug.
+Retain the test when the defect could realistically recur and its consequences justify the ongoing cost. Prefer the narrowest durable test of the violated contract or invariant.
 
-Not every historical bug requires a permanent test. If the underlying design has made recurrence impossible, the test has become redundant, or a stronger invariant test now subsumes it, remove or consolidate it.
+Remove or consolidate a test when the underlying design has made recurrence impossible, its protection is redundant, or a stronger invariant test subsumes it.
 
 ## Deleting or rewriting tests
 
-Tests are maintained code, not permanent historical artifacts.
-
-Delete or rewrite a test when its maintenance cost exceeds the confidence it provides.
+Delete or rewrite tests when maintenance cost exceeds their continuing
+protective value.
 
 Strong candidates include tests that:
 
 - require updates after harmless internal changes;
-- directly assert private constants without protecting a real contract;
-- duplicate the same policy across several layers;
+- assert incidental private constants or snapshot large amounts of incidental output;
 - verify mock choreography rather than outcomes;
-- snapshot large amounts of incidental output;
-- test trivial private helpers already exercised through stable package behavior;
-- protect risks already covered more effectively elsewhere;
+- duplicate policy across layers, retest trivial helpers covered by package
+  behavior, or protect risks covered more effectively elsewhere;
 - are flaky, misleading, obsolete, or disproportionately expensive to diagnose; or
 - no longer correspond to a plausible failure mode.
 
-Several brittle tests may encode one genuine requirement. Replace them with one durable behavior-level or invariant test rather than preserving all of them.
-
-Deleting a low-value test can improve the quality of the suite by reducing noise, maintenance burden, and friction around legitimate change.
+Several brittle tests may encode one genuine requirement. Consolidate them into one durable behavior-level or invariant test.
 
 ## Reviewing a proposed test
 
-Use the following questions when the value, boundary, or durability of a proposed test is not self-evident. Significant test additions should be reviewable against them, but written answers are not required for every routine test.
+Use the following questions when a proposed test needs closer evaluation. Significant test additions should be reviewable against them; written answers are optional for routine tests.
 
-1. What realistic defect would it catch?
-2. How likely is that defect?
-3. How consequential would it be?
-4. Is the behavior already protected elsewhere?
-5. At which layer should this behavior be owned?
-6. Does the test assert a durable contract or an incidental implementation detail?
-7. Could the implementation be refactored without changing the behavior and without editing this test?
-8. What should cause this test to fail?
-9. What legitimate changes should not cause this test to fail?
-10. What ongoing maintenance, execution, and diagnostic cost will the test impose?
-11. Is there a smaller or more direct test that protects the same risk?
-
-Do not add the test when its expected lifetime cost exceeds its expected protective value.
-
-When deciding not to test plausible behavior, record or be able to explain why the risk is low, already protected, obvious, reversible, or cheaper to detect elsewhere.
+1. Which requested behavior or established contract does it protect, and what realistic defect would violate it?
+2. How likely and consequential is that defect, and is it already protected elsewhere?
+3. What is the narrowest stable boundary that can test it clearly?
+4. Would a correct internal refactor or a change to an incidental default require test edits?
+5. Does its confidence justify its maintenance, execution, and diagnostic cost, or would a smaller test suffice?
 
 ## Definition of sufficient
 
 A test suite is sufficient when:
 
-- important contracts and invariants are protected;
-- meaningful boundaries and failure modes are exercised;
-- realistic and consequential regressions are credibly protected against silent recurrence;
-- behavior involving data integrity, destructive operations, compatibility, security, concurrency, idempotency, and recovery is credibly protected;
-- important external boundaries have realistic integration coverage;
-- representative complete workflows are tested;
-- failures provide useful signal rather than redundant noise;
-- legitimate internal changes usually do not require test edits; and
+- important requested behavior and established contracts are protected against realistic, consequential failures;
+- boundaries and component interactions have coverage where they introduce meaningful risks not protected elsewhere;
+- failures provide useful, distinct signals;
+- tests usually remain stable through legitimate internal changes; and
 - additional tests would mostly repeat existing protection or preserve inconsequential implementation details.
 
-Sufficiency is a risk judgment, not a coverage percentage or test count.
-Reassess it as the library, its consumers, and the consequences of failure
-evolve.
-
-The governing rule is:
-
-> Test heavily where failure is consequential, subtle, or difficult to detect after the fact. Test lightly where failure is obvious, reversible, and inexpensive—and retain no test whose lifetime cost exceeds the confidence it provides.
+Reassess sufficiency as the application, its users, and the consequences of
+failure evolve.

docs/release.md

@@ -1,242 +1,159 @@
 # Release Procedure
 
-## Release Model
+This runbook implements the [release policy](policy/release.md). Promptkit
+publishes source-only Go module tags from `main` to `origin`, followed by a
+page-only Gitea release. Successful publication of the intended tag, with its
+source commit available on `main`, completes the source release. Report hosted
+delivery separately. Do not wait for CI or Go module availability.
 
-Promptkit publishes a Go library through source commits and semantic Go module
-tags. It does not publish runnable binaries or binary packages and does not
-currently use hosted CI. The release maintainer performs and records the
-required validation.
+## Prepare The Candidate
 
-`v0.1.0` is the initial published release. Later releases use semantic
-`vMAJOR.MINOR.PATCH` tags. Before `v1`, minor releases may change the public
-API and patch releases preserve compatibility within their minor line. Every
-pre-`v1` release note must summarize compatibility, identify public API
-changes, and state any action required of consumers.
+Required tools are Git, Go with race-detector support, and POSIX shell. Work
+from the repository root with push access to `origin`.
 
-Promptkit releases are source-only. The annotated tag message is the release
-note; there is no separate hosted release or binary packaging step.
-
-## Establish The Candidate
-
-Choose a version that has not been published and export it as
-`RELEASE_VERSION`. Run every command in this procedure from the Promptkit
-repository root in the same POSIX shell. Do not reuse `v0.1.0` or another
-existing version.
-
-The following guard derives the release commit from `HEAD` and stops on a
-missing or malformed version, a checkout other than synchronized `main`,
-uncommitted changes, an active Go workspace, a module replacement, a vendor
-tree, or an existing local or remote tag:
+Use the latest Go installation available locally for the release tool and its
+child commands. Put it first on `PATH` and keep it selected throughout:
 
 ```sh
-set -eu
-
-: "${RELEASE_VERSION:?export an unpublished vMAJOR.MINOR.PATCH version}"
-if ! printf '%s\n' "$RELEASE_VERSION" |
-    grep -Eq '^v(0|[1-9][0-9]*)\.(0|[1-9][0-9]*)\.(0|[1-9][0-9]*)$'
-then
-    printf '%s\n' "invalid release version: $RELEASE_VERSION" >&2
-    exit 1
-fi
-
-RELEASE_COMMIT=$(git rev-parse --verify 'HEAD^{commit}')
-export RELEASE_COMMIT
-
-check_release_candidate() {
-    test "$(git branch --show-current)" = main
-    test -z "$(git status --porcelain)"
-
-    gowork_value=$(go env GOWORK)
-    case "$gowork_value" in
-        ''|off) ;;
-        *)
-            printf '%s\n' "active Go workspace: $gowork_value" >&2
-            return 1
-            ;;
-    esac
-
-    test -z "$(git ls-files go.work go.work.sum)"
-    test ! -e vendor
-    if grep -Eq '^[[:space:]]*replace([[:space:]]|\()' go.mod
-    then
-        printf '%s\n' 'go.mod contains a replacement' >&2
-        return 1
-    fi
-
-    git fetch origin main --tags
-    test "$RELEASE_COMMIT" = \
-        "$(git rev-parse --verify 'refs/remotes/origin/main^{commit}')"
-
-    if git show-ref --verify --quiet "refs/tags/$RELEASE_VERSION"
-    then
-        printf '%s\n' "local tag already exists: $RELEASE_VERSION" >&2
-        return 1
-    fi
-    if test -n "$(
-        git ls-remote --tags origin \
-            "refs/tags/$RELEASE_VERSION" \
-            "refs/tags/$RELEASE_VERSION^{}"
-    )"
-    then
-        printf '%s\n' "remote tag already exists: $RELEASE_VERSION" >&2
-        return 1
-    fi
-}
-
-check_release_candidate
+export GOTOOLCHAIN=local
+go version
 ```
 
-Do not continue unless the guard completes successfully. In particular, push
-the intended commit through the normal `main` branch workflow before release;
-the tag procedure is not a substitute for publishing the source commit.
-
-## Validate The Candidate
-
-Confirm the module and root package metadata:
+For a minor release, normally advance the consumer minimum to the initial
+stable patch of that Go series:
 
 ```sh
-go list -m -f '{{.Path}} {{.GoVersion}}'
-go list -f '{{.Name}} {{.ImportPath}}' .
+release_go_version=$(go env GOVERSION)
+release_go_series=${release_go_version#go}
+release_go_series=${release_go_series%.*}
+go mod edit "-go=${release_go_series}.0"
 ```
 
-The output must be:
-
-```text
-gitea.maximumdirect.net/eric/promptkit 1.25.5
-promptkit gitea.maximumdirect.net/eric/promptkit
-```
-
-As a release prerequisite, run the complete
-[maintainer validation workflow](development.md#maintainer-validation) against
-the clean candidate. Do not substitute a partial command list: the development
-guide owns the tests, race checks, analysis, build, both offline examples,
-formatting, Markdown links, generated-output and credential review, and
-repository hygiene. Record the successful workflow result with the candidate.
-
-## Write The Release Note
-
-Prepare a plain-text annotated-tag message outside the repository and export
-its path as `RELEASE_NOTES_FILE`. Use this form, replacing each summary with
-release-specific text; write `None.` when there are no public API changes or
-consumer actions:
-
-```text
-Promptkit vMAJOR.MINOR.PATCH
-
-Validated commit: full commit ID
-Compatibility: compatibility summary
-Public API changes: changes or None.
-Consumer action: required action or None.
-```
-
-After writing it, require all release-note fields, the selected version, and
-the validated commit to be present:
+Retain a higher patch minimum when required by a dependency or identified fix.
+For a patch release, preserve the existing minimum. Explain minimum increases
+under the [versioning policy](policy/release.md#versioning-and-compatibility).
+Prepare dependency caches before committing the candidate:
 
 ```sh
-: "${RELEASE_NOTES_FILE:?export the path to the release-note file}"
-test -f "$RELEASE_NOTES_FILE"
-test -s "$RELEASE_NOTES_FILE"
-grep -F "Promptkit $RELEASE_VERSION" "$RELEASE_NOTES_FILE"
-grep -F "Validated commit: $RELEASE_COMMIT" "$RELEASE_NOTES_FILE"
-grep -F 'Compatibility:' "$RELEASE_NOTES_FILE"
-grep -F 'Public API changes:' "$RELEASE_NOTES_FILE"
-grep -F 'Consumer action:' "$RELEASE_NOTES_FILE"
+go mod download
 ```
 
-Inspect the complete message and confirm that it accurately records the
-compatibility impact, public API changes, and required consumer action.
+Choose an unused stable version and create `docs/releases/<version>.md` with a
+title containing `Promptkit` and that exact version, a short summary, and any
+compatibility impact or upgrade instructions. End the note with:
 
-## Create And Inspect The Tag
-
-Run the candidate guard again immediately before tag creation. This ensures
-that validation or release-note preparation did not change the checkout and
-that the commit is still published and untagged:
-
-```sh
-check_release_candidate
+```markdown
+[COPYRIGHT](https://forge.maximumdirect.net/base-layer/promptkit/src/tag/vX.Y.Z/COPYRIGHT) | [LICENSE](https://forge.maximumdirect.net/base-layer/promptkit/src/tag/vX.Y.Z/LICENSE) | [ATTRIBUTION](https://forge.maximumdirect.net/base-layer/promptkit/src/tag/vX.Y.Z/docs/oss/attribution.md)
 ```
 
-Create the annotated tag from the prepared release note and bind it explicitly
-to the validated commit:
+Substitute the selected version and check the paths against the candidate;
+future tag URLs need not resolve during preparation. Follow the
+[OSS policy](policy/oss.md). Historical notes and tags remain unchanged.
+
+Supply delivery credentials through `RELEASEKIT_GITEA_TOKEN` or the ignored
+local file `.releasekit/gitea-token`, with owner-only file permissions. A
+nonblank environment value takes precedence. Source publication neither reads
+nor requires this token. The [profile](../releasekit.json) selects the token
+path for committed-file protection; keep credentials out of commits and logs.
+
+Retire completed roadmaps, preserve unfinished work, and repair affected links.
+Review and commit the candidate on `main`, including any module updates, then
+push `main` to `origin` before publishing its tag.
+
+## Validate And Publish
 
 ```sh
-git tag --annotate "$RELEASE_VERSION" \
-    --file "$RELEASE_NOTES_FILE" \
-    "$RELEASE_COMMIT"
+./scripts/release.sh vMAJOR.MINOR.PATCH
 ```
 
-Inspect both the tag message and its source commit before publication:
+The launcher uses `go tool` to build or reuse the module-selected ReleaseKit
+executable in Go's cache, using the selected local toolchain. ReleaseKit runs
+the [profile's checks](../releasekit.json) once, verifies publication state, and
+publishes only the intended tag. Preserve its native exit status and diagnostic.
+Successful publication needs no separate remote read-back or CI wait.
+
+For optional local diagnosis, use `./scripts/releasekit.sh check VERSION`.
+It is not a required step before publication. The upstream
+[candidate-check guide](https://forge.maximumdirect.net/base-layer/releasekit/src/tag/v0.9.0/docs/consumers/check.md)
+and [CLI contract](https://forge.maximumdirect.net/base-layer/releasekit/src/tag/v0.9.0/docs/cli.md)
+own shared checks and exit-status definitions.
+
+## Deliver The Hosted Release
+
+After source publication, keep a clean checkout at the published tag's commit:
 
 ```sh
-test "$(git cat-file -t "refs/tags/$RELEASE_VERSION")" = tag
-git show --no-patch --decorate "refs/tags/$RELEASE_VERSION"
-test "$(
-    git rev-parse --verify "refs/tags/$RELEASE_VERSION^{commit}"
-)" = "$RELEASE_COMMIT"
+./scripts/releasekit.sh deliver vMAJOR.MINOR.PATCH
 ```
 
-If inspection finds an error, delete the unpublished local tag, correct the
-release note or candidate, and repeat the guards. Never move or recreate a tag
-that has been published.
+The profile selects page-only Gitea delivery for this repository. The hosted
+body is the complete checked-in note, including its licensing footer. There are
+no binary assets, checksum manifests, candidate revalidation, or CI waits.
+ReleaseKit checks delivery credentials during delivery, not as a publication
+gate. Report the hosted URL on success; delivery failure leaves the source
+release complete.
 
-## Publish The Selected Tag
-
-Push only the selected tag ref. Do not use `git push --tags`:
+For a later retry, use a clean checkout of the original tag with its recorded
+module files and supply the token there. In a separate checkout, for example:
 
 ```sh
-git push origin \
-    "refs/tags/$RELEASE_VERSION:refs/tags/$RELEASE_VERSION"
-```
-
-## Verify Publication
-
-Compare the remote annotated-tag object with the local object, then compare the
-remote peeled source commit with the validated commit:
-
-```sh
-remote_tag=$(
-    git ls-remote --tags origin "refs/tags/$RELEASE_VERSION" |
-        awk 'NR == 1 { print $1 }'
-)
-remote_commit=$(
-    git ls-remote --tags origin "refs/tags/$RELEASE_VERSION^{}" |
-        awk 'NR == 1 { print $1 }'
-)
-test -n "$remote_tag"
-test "$remote_tag" = \
-    "$(git rev-parse --verify "refs/tags/$RELEASE_VERSION")"
-test "$remote_commit" = "$RELEASE_COMMIT"
-```
-
-Finally, resolve the version as an ordinary Go module in a temporary module
-outside this repository and without a workspace or replacement:
-
-```sh
-resolution_dir=$(mktemp -d)
 (
-    trap 'rm -rf "$resolution_dir"' 0 1 2 15
-    cd "$resolution_dir"
-    GOWORK=off go mod init example.com/promptkit-release-check
-    GOWORK=off go mod download \
-        "gitea.maximumdirect.net/eric/promptkit@$RELEASE_VERSION"
-    resolved_version=$(
-        GOWORK=off go list -m -f '{{.Version}}' \
-            "gitea.maximumdirect.net/eric/promptkit@$RELEASE_VERSION"
-    )
-    test "$resolved_version" = "$RELEASE_VERSION"
+  set -e
+  release_version=vMAJOR.MINOR.PATCH
+  release_tag_ref="refs/tags/$release_version"
+  git fetch origin "$release_tag_ref:$release_tag_ref"
+  release_tag_commit=$(git rev-parse --verify "$release_tag_ref^{commit}")
+  git switch --detach "$release_tag_commit"
+  release_worktree_status=$(git status --porcelain --untracked-files=all)
+  if [ -n "$release_worktree_status" ]; then
+    printf '%s\n' 'Delivery retry requires a clean tagged checkout.' >&2
+    exit 1
+  fi
+  ./scripts/releasekit.sh deliver "$release_version"
 )
 ```
 
-Promptkit must publish and verify the required version before Scriptorium or
-another consumer publishes a release that depends on it. This ordering does
-not replace the consumer project's own release procedure. Released consumers
-must select the published Promptkit tag through ordinary module resolution,
-without a workspace, replacement, vendored Promptkit source, or unpublished
-revision.
+Resolve uncertain hosted writes before retrying the same immutable version.
+Matching content is accepted; conflicts are not overwritten. Do not rerun
+`publish` or move tags to repair delivery. Follow the upstream
+[delivery guide](https://forge.maximumdirect.net/base-layer/releasekit/src/tag/v0.9.0/docs/consumers/delivery.md)
+for result interpretation and recovery. These commands apply to tags containing
+this delivery configuration; older releases retain their original workflows.
 
-## Policy Changes
+## Failure Recovery
 
-Document and approve a durable policy change before introducing hosted
-automation, binary artifacts, or different release governance. Update this
-procedure in the same change so maintainers do not rely on hidden release
-requirements.
+For source-publication failures, use ReleaseKit's status and diagnostic to
+distinguish a candidate defect from a publication conflict. Correct and commit
+candidate defects, push `main`, and rerun publication as appropriate. A failure
+after local tag creation may leave a ref even when its object ID is unavailable.
+
+Status 3 means the push outcome remains unknown. Do not retry or remove the local
+tag until the authoritative outcome is resolved. ReleaseKit attempts to save
+private publication evidence and reports a record path or a separate recording
+warning. A recording failure does not change the publication outcome.
+
+Use the recovery command reported for that attempt, or discover local records
+and explicitly select the matching attempt:
+
+```sh
+./scripts/releasekit.sh publication-attempts
+./scripts/releasekit.sh reconcile --record /absolute/path/to/selected-record.json
+```
+
+Discovery is local; reconciliation observes the original destination without
+rerunning validation, changing tags, or delivering hosted content. Neither is
+a routine gate after successful publication. Follow the upstream
+[recovery procedure](https://forge.maximumdirect.net/base-layer/releasekit/src/tag/v0.9.0/docs/release.md#resolve-publication-failures)
+for outcomes, incomplete evidence, and older attempts without records.
+
+Records are stored under Git's common directory by default and can contain
+credential-bearing endpoints. Keep them private, out of source control and
+shared logs; retain them while needed for recovery and remove them when that
+value expires. See the [evidence guide](https://forge.maximumdirect.net/base-layer/releasekit/src/tag/v0.9.0/docs/consumers/publish.md#retain-and-reconcile-publication-evidence)
+for storage and retention details.
+
+Current absence does not prove a version was never published. Remove a retained
+local tag only after establishing absence and confirming it was not already
+published or used by another release channel. Never reuse a known published
+version or replace a published tag or hosted page. Corrections require a new
+appropriate version.

docs/releases/v0.10.0.md

@@ -0,0 +1,44 @@
+# Promptkit v0.10.0
+
+This supplemental guide summarizes the `v0.10.0` changes and migration needs.
+The annotated tag remains the authoritative release record.
+
+## Transition Summary
+
+`v0.10.0` adopts independently versioned OpenRouter and Rakestrawhome catalog
+modules whose external profile data can change between catalog releases. It
+also introduces the maintained literal names
+`promptkit-openrouter-standard` and `promptkit-openrouter-auto`.
+
+The public package now delegates assembly to an internal engine while retaining
+the existing Go API. This refactor requires no consumer code changes.
+
+This release selects OpenRouter `v2.20260922.1` and Rakestrawhome
+`v2.20260922.0`; their exact inventories are linked from the
+[migration identity mapping](../consumers/configuration-migration.md#identity-mapping).
+
+The maintained names provide continuity only within the catalog versions
+selected and tested by a Promptkit release and when consumers do not override
+the name or its base. Their current selections may change. External catalog
+profile IDs can be removed, renamed, or modified, so an application that
+selects one directly may need to update its configuration when upgrading.
+
+## Consumer Migration
+
+Before upgrading, review every profile ID used by prompts, requests, and
+application-owned inheritance. Choose a maintained name, accept an external
+catalog profile's volatility, or define a complete application-owned profile
+when the application must control its target and settings. An alias that
+inherits from an external base is not a frozen copy.
+
+In particular, the selected OpenRouter catalog removes `deepseek-3-2`,
+`deepseek-4-flash`, and `deepseek-4-pro`. Consumers using those IDs must choose
+an available profile or supply their own complete definition; the new
+`deepseek-flash-latest` and `deepseek-pro-latest` profiles are not automatic
+aliases and need not preserve the previous settings.
+
+See the [release policy](../policy/release.md#versioning-and-compatibility) for the
+compatibility boundary, the [migration identity mapping](../consumers/configuration-migration.md#identity-mapping)
+for current resource mappings, and
+the [consumer guide](../consumers/pkg-promptkit.md#choose-a-profile-ownership-model)
+for configuration choices.

docs/releases/v0.11.0.md

@@ -0,0 +1,76 @@
+# Promptkit v0.11.0
+
+This release adds reliable, observable model execution and changes Promptkit's
+license to LGPL-3.0-or-later.
+
+## Summary
+
+- Built-in model calls can recover from selected transient HTTP failures.
+- Optional execution observations report attempts, timing, outcomes, and known
+  token usage, including work completed before a run fails.
+- Public errors distinguish transport failures, malformed responses, and retry
+  exhaustion without parsing diagnostic text.
+
+## Compatibility
+
+The built-in client now defaults to two additional transient attempts. This can
+increase latency, request counts, and cost; retries do not guarantee exactly-once
+generation. Injected `LLMClient` implementations retain their own retry policy
+and are not automatically replayed. Existing generation timeouts apply per
+attempt, while the caller context governs the complete operation. See the
+[provider integration contract](../integrations/openai-compatible-chat.md).
+
+Public retry and observation controls are additive. Existing successful-result
+usage semantics remain unchanged, and operational failures still return no
+partial `RunResult`. Observations distinguish unknown usage from reported zero;
+they are not a complete billing ledger. See the
+[consumer guide](../consumers/pkg-promptkit.md) and
+[observation GoDoc](../../observations.go).
+
+Promptkit is now licensed under LGPL-3.0-or-later. Review the
+[license summary](../../README.md#license), [LGPL text](../../LICENSE), and
+[accompanying GPL text](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/COPYING) before adoption or distribution.
+Previously published releases and dependency licenses are unchanged.
+
+## Upgrade
+
+Update the dependency:
+
+```sh
+go get gitea.maximumdirect.net/eric/promptkit@v0.11.0
+```
+
+To retain single-attempt behavior for built-in generation, set
+`Config.TransientRetries` to a pointer to zero. `RunRequest.TransientRetries`
+can override the engine choice per request. Review any existing application or
+HTTP-transport retry loops to avoid unintentionally multiplying attempts.
+Transient retries and structural repairs have separate budgets; provide a
+caller deadline when a total execution time limit is needed.
+
+Observation is opt-in through `RunRequest.Observer`, including for prepared
+execution. Callbacks are synchronous; shared observers must support concurrent
+runs. Injected clients can supply `GenerateResponse.UsagePresence` to explicitly
+report known zero counts. No observer or usage-presence changes are required
+for existing consumers. See the [consumer guide](../consumers/pkg-promptkit.md)
+for adoption and the [offline example](../../examples/go-library/observe/main.go)
+for a structured proposal-and-assessment workflow.
+
+## Changes
+
+- Centralized built-in retry coordination with cancellable backoff and provider
+  retry timing. Retries and structural repair share existing backend capacity;
+  backoff releases active permits without releasing logical run admission.
+- Added `ErrTransport`, `ErrMalformedResponse`, `ErrRetriesExhausted`, and
+  `RetryExhaustedError`, retaining applicable underlying error identities.
+- Added request-scoped lifecycle observations and presence-aware token accounting
+  for ordinary and prepared execution, without recording prompt/response bodies.
+- Isolated delegated operations' observation state and made injected-client
+  attempt classifications consistent with terminal classifications.
+- Added an offline observation example and regression coverage for retry,
+  repair, capacity, error, and observation interactions.
+- Adopted LGPL-3.0-or-later and streamlined the documented local release process.
+
+Backend catalog selections are unchanged from v0.10.0: OpenRouter
+`v2.20260922.1` and Rakestrawhome `v2.20260922.0`. See the
+[release compatibility policy](../policy/release.md#versioning-and-compatibility)
+and [migration identity mapping](../consumers/configuration-migration.md#identity-mapping).

docs/releases/v0.11.1.md

@@ -0,0 +1,43 @@
+# Promptkit v0.11.1
+
+This patch updates the maintained catalogs to MIT-licensed releases and
+improves licensing documentation.
+
+## Summary
+
+Both OpenRouter and Rakestrawhome catalogs now use `v2.20260926.0`. These
+catalog releases change licensing and documentation only; their profiles,
+backend definitions, and Go code are unchanged.
+
+## Compatibility
+
+No public API, runtime behavior, or profile changes are introduced.
+Promptkit remains LGPL-3.0-or-later; the catalogs are now MIT-licensed.
+See the [project grant](../../COPYRIGHT),
+[component attribution](../oss/attribution.md), and
+[release compatibility policy](../policy/release.md#versioning-and-compatibility).
+
+## Upgrade
+
+Update the module dependency:
+
+```sh
+go get gitea.maximumdirect.net/eric/promptkit@v0.11.1
+```
+
+No code or configuration migration is required from v0.11.0.
+Consumers upgrading from earlier versions should also read the
+[v0.11.0 migration guide](v0.11.0.md).
+
+## Changes
+
+- Adopt the MIT-licensed catalog releases and update inventory links.
+- Add centralized attribution and dependency license texts.
+- Separate the project grant into COPYRIGHT and consolidate the LGPLv3 and
+  incorporated GPLv3 texts in LICENSE, replacing the separate COPYING file.
+- Align documentation ownership and future release navigation with the
+  [OSS policy](../policy/oss.md), without adding release gates.
+- Repair the historical v0.11.0 GPL-text link and retire the completed
+  licensing roadmap.
+
+[COPYRIGHT](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.1/COPYRIGHT) | [LICENSE](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.1/LICENSE) | [ATTRIBUTION](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.1/docs/oss/attribution.md)

docs/releases/v0.11.2.md

@@ -0,0 +1,38 @@
+# Promptkit v0.11.2
+
+This patch aligns repository policies with the current standard and clarifies
+licensing and attribution for the copied policy documents.
+
+## Summary
+
+The update adds policy provenance, selective contributor reading guidance, and
+consistent upstream policy language while preserving Promptkit-specific rules.
+
+## Compatibility
+
+No public API, runtime behavior, dependency, or catalog changes are introduced.
+Promptkit remains LGPL-3.0-or-later. The [project grant](../../COPYRIGHT)
+identifies the separate CC-BY-SA-4.0 scope for copied or adapted policy material;
+the [attribution notices](../oss/attribution.md) identify the affected documents.
+The existing release checks and successful-tag-push completion point are unchanged.
+
+## Upgrade
+
+```sh
+go get gitea.maximumdirect.net/eric/promptkit@v0.11.2
+```
+
+No code or configuration migration is required from v0.11.1.
+
+## Changes
+
+- Align adopted policies with Standard Repository Policies while retaining
+  library-specific requirements and deliberate local decisions.
+- Add per-document upstream source links and the
+  [policy-set provenance document](../policy/meta.md).
+- Add selective policy reading to the [contributor guide](../development.md)
+  and update agent orientation.
+- Standardize factual attribution entries and identify copied-policy licensing.
+- Retire the completed policy-alignment roadmap.
+
+[COPYRIGHT](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.2/COPYRIGHT) | [LICENSE](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.2/LICENSE) | [ATTRIBUTION](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.2/docs/oss/attribution.md)

docs/releases/v0.12.0.md

@@ -0,0 +1,74 @@
+# Promptkit v0.12.0
+
+This release replaces Promptkit's configuration loader with ConfigKit and
+migrates maintained catalogs to keyed `/v3` assets. It intentionally changes
+the public construction API and configuration format.
+
+## Summary
+
+Backends, profiles, and versioned prompts now share one layered configuration
+snapshot built when `NewEngine` is called. ConfigKit supplies composition,
+inheritance, presence, and source history; Promptkit validates the final
+effective definitions and captures prompt content and templates. Consumers can
+inspect the captured configuration explicitly without triggering a reload.
+
+## Compatibility
+
+This is an incompatible minor release under Promptkit's pre-1.0
+[versioning policy](../policy/release.md#versioning-and-compatibility). The
+former source options, `Config.PromptDir`, `Config.ProfileDir`, and the old
+profile/backend convenience values have been removed. Configuration documents
+now use literal keys beneath `backends`, `profiles`, and `prompts`; embedded
+`id`/`version` fields and `base_profile` are no longer accepted. Source groups
+compose strictly within a group and overlay in the supplied order. Explicit
+zero, false, empty, and null values can now differ from omission; maps overlay
+recursively, and an empty map does not clear an inherited map.
+
+All final definitions, including unused ones, are validated during engine
+construction. Prompt text and static templates are captured then; an existing
+engine does not automatically reload changed configuration or message files.
+Request artifacts, schemas, credentials, and provider operations retain their
+runtime boundaries. See the [format reference](../formats.md) for the current
+configuration contract and the [consumer migration guide](../consumers/pkg-promptkit.md#migrate-from-the-previous-configuration-api)
+for API and source-selection changes.
+
+Maintained OpenRouter and Rakestrawhome catalogs are selected as `/v3` modules,
+both at `v3.20260930.0`. Their asset format changed; consumers that read catalog
+assets directly must adopt the keyed format. The maintained Promptkit profile
+IDs remain available under the [release compatibility policy](../policy/release.md#versioning-and-compatibility),
+which does not promise fixed profile data.
+
+## Upgrade
+
+```sh
+go get gitea.maximumdirect.net/eric/promptkit@v0.12.0
+```
+
+Replace removed source and backend registration options with
+`Config.Sources`, explicitly ordered ConfigKit underlays, and optionally a
+highest-priority main input. Use `DirectoryInputs` or `FilesystemInputs` for
+selected directories, `ProfileSource` and `BackendSource` for typed
+programmatic values, or `configkit.Mapping` for explicit null and combined
+documents. Move profile `base_profile` to `extends` and embedded identities to
+mapping keys. Review zero/false/null values and any assumptions that replacing
+a model or backend discards unrelated explicit settings. Rebuild the engine
+when definitions or referenced message files change. The
+[consumer guide](../consumers/pkg-promptkit.md#migrate-from-the-previous-configuration-api)
+provides the migration details and maintained examples.
+
+## Changes
+
+- Add ConfigKit v0.5.0 as the configuration boundary and consume released
+  OpenRouter and Rakestrawhome `/v3` catalog modules.
+- Add keyed backend, profile, and prompt definitions with final-value domain
+  validation, explicit presence/null behavior, and source-relative content
+  capture.
+- Add `ConfigurationSources`, directory discovery helpers, programmatic
+  profile/backend source builders, and read-only configuration inspection.
+- Preserve execution workflows while rendering captured templates, applying
+  backend defaults after inheritance, and honoring explicit suppression of
+  provider settings.
+- Remove superseded loader, catalog preflight, and inheritance machinery;
+  update examples, format guidance, and internal documentation.
+
+[COPYRIGHT](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.12.0/COPYRIGHT) | [LICENSE](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.12.0/LICENSE) | [ATTRIBUTION](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.12.0/docs/oss/attribution.md)

docs/releases/v0.13.0.md

@@ -0,0 +1,69 @@
+# Promptkit v0.13.0
+
+This release adds prompt requirement discovery and profile/prompt listing,
+reduces avoidable configuration and execution allocations, and standardizes
+maintainer publication through ReleaseKit. The minimum supported Go version
+increases to 1.26.0.
+
+## Discover Requirements And Available Definitions
+
+- `Engine.InspectRequirements` reports recognizable input and caller-variable
+  dependencies for a selected prompt and supplied caller context before input
+  artifacts exist. It classifies dependencies as required, conditional, or
+  optional and reports structured limitations when discovery is incomplete.
+  Discovery coverage is not proof that requirements are satisfied or execution
+  will succeed. See the [requirement inspection guide](../consumers/pkg-promptkit.md#inspect-requirements-before-preparation)
+  and [GoDoc](../../requirements.go).
+- `Engine.ListProfiles` returns effective profile IDs, and `Engine.ListPrompts`
+  returns exact ID/version pairs as `PromptIdentity` values. Lists come from the
+  engine's captured configuration, are lexically ordered and caller-owned, and
+  include consumer definitions and maintained profiles that remain effective.
+  Listing performs no provider or execution work. See the
+  [discovery guide](../consumers/pkg-promptkit.md#discover-available-resources-and-collections)
+  and [GoDoc](../../listing.go).
+
+## Efficiency And Tooling
+
+- Reduce response-sized copies in basic validation and rendered-message output,
+  and encode valid provider extra parameters only as part of the complete
+  request rather than serializing them twice.
+- Use bounded initial file-size reservations without limiting accepted content,
+  releasing excess backing storage for small or empty completed artifacts.
+- Reuse owned preparation values when creating prepared-execution snapshots,
+  while preserving independent inspection details and caller-owned data.
+- Use lightweight captured profile identities and prompt metadata for inspection
+  instead of copying execution targets and definitions unnecessarily.
+- Upgrade ConfigKit to v0.6.0 for more efficient loading, resolution, metadata,
+  and inspection processing. Configuration semantics remain unchanged; these
+  improvements primarily benefit construction and explicit operator inspection,
+  not upstream inference latency.
+- Adopt module-selected ReleaseKit v0.3.0 for maintainer validation and tag
+  publication. The [release procedure](../release.md) owns the current commands
+  and recovery workflow; successful tag publication remains the completion
+  point, without waiting for CI.
+
+These changes preserve existing execution and ownership contracts. Benefits
+depend on workload; this release introduces no fixed performance guarantee.
+
+## Compatibility And Upgrade
+
+Consumers upgrading from v0.12.0 need **Go 1.26.0 or newer**, rather than the
+previous 1.25.5 minimum. Update application build environments and any CI images
+still using Go 1.25 before upgrading:
+
+```sh
+go get gitea.maximumdirect.net/eric/promptkit@v0.13.0
+```
+
+No application-code or configuration migration is required from v0.12.0 beyond
+the Go toolchain update. The new discovery methods are optional additions;
+existing construction, inspection, preparation, and execution workflows remain
+available. See [Go compatibility](../consumers/pkg-promptkit.md#go-compatibility)
+for the distinction between the consumer minimum and maintainer toolchain.
+
+Maintainers should follow the current release runbook rather than the retired
+`scripts/release-check.sh`. The publication entry point remains
+`scripts/release.sh`; ReleaseKit now owns its shared validation and publication
+mechanics.
+
+[COPYRIGHT](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.13.0/COPYRIGHT) | [LICENSE](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.13.0/LICENSE) | [ATTRIBUTION](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.13.0/docs/oss/attribution.md)

docs/releases/v0.14.0.md

@@ -0,0 +1,48 @@
+# Promptkit v0.14.0
+
+This release adds comparison of captured configuration inspections and updates
+ConfigKit and the maintainer release tool. The minimum supported Go version
+remains 1.26.0.
+
+## Compare Captured Configurations
+
+`ConfigurationInspection.Diff` compares the effective configuration captured by
+two engines. Consumers can use it to review a candidate configuration, inspect
+an overlay's effect, or present configuration changes to operators.
+
+Comparison delegates to ConfigKit and returns its difference records, including
+presence information, paired redaction, and protected-change indicators. It
+does not reload sources or contact providers. Source-history differences alone
+do not produce value differences.
+
+An empty diff is not proof of equivalent execution: referenced file contents,
+resolved credentials, later execution preparation, and request-specific
+overrides are outside the comparison. Records can contain ordinary application
+data; callers retain responsibility for their disclosure.
+
+See the [configuration comparison guide](../consumers/pkg-promptkit.md#inspect-the-captured-configuration)
+and [GoDoc](../../configuration_inspection.go) for usage and the exact contract.
+
+## Dependency And Tooling Updates
+
+- Upgrade ConfigKit from v0.6.0 to v0.7.1, providing the upstream comparison API.
+  v0.7.1 also updates ConfigKit's maintainer tooling without changing its runtime
+  behavior from v0.7.0.
+- Upgrade the module-selected ReleaseKit tool from v0.3.0 to v0.4.1. Promptkit
+  retains its existing root-module profile, publication entry point, and release
+  gates. See the [release procedure](../release.md) for maintainer commands.
+
+## Compatibility And Upgrade
+
+No application-code, configuration, or Go-toolchain migration is required from
+v0.13.0. Configuration comparison is an optional API addition; existing
+construction, inspection, preparation, and execution workflows remain available.
+
+```sh
+go get gitea.maximumdirect.net/eric/promptkit@v0.14.0
+```
+
+See [Go compatibility](../consumers/pkg-promptkit.md#go-compatibility) for the
+distinction between the consumer minimum and maintainer toolchain.
+
+[COPYRIGHT](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.14.0/COPYRIGHT) | [LICENSE](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.14.0/LICENSE) | [ATTRIBUTION](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.14.0/docs/oss/attribution.md)

docs/releases/v0.15.0.md

@@ -0,0 +1,34 @@
+# Promptkit v0.15.0
+
+This release adds error-only validation of a concrete prompt request without
+calling a model. The minimum supported Go version remains 1.26.0.
+
+## Validate A Concrete Request
+
+`Engine.ValidateRequest` accepts an ordinary `RunRequest` and returns nil when
+it passes preparation, or the existing preparation error. Consumers can use it
+for fail-fast checks, including downstream CLI validation commands, without
+receiving a prepared snapshot.
+
+The method shares `Prepare`'s request conversion and preparation path. Unlike
+requirements inspection, it checks the actual supplied values and bindings.
+Artifact and schema resource I/O still occurs; success does not certify provider
+acceptance or later execution success. A subsequent run prepares independently.
+Validation is optional, not a required step before preparation or execution.
+
+See the [request validation guide](../consumers/pkg-promptkit.md#validate-a-concrete-request)
+and [GoDoc](../../engine.go) for usage and the exact contract. Consumers that
+need execution tied to a checked snapshot can use the existing
+[prepared execution workflow](../consumers/pkg-promptkit.md#prepare-now-and-execute-the-same-snapshot-later).
+
+## Compatibility And Upgrade
+
+No application-code, configuration, or Go-toolchain migration is required from
+v0.14.0. This is an optional API addition; existing inspection, preparation, and
+execution workflows remain available.
+
+```sh
+go get gitea.maximumdirect.net/eric/promptkit@v0.15.0
+```
+
+[COPYRIGHT](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.15.0/COPYRIGHT) | [LICENSE](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.15.0/LICENSE) | [ATTRIBUTION](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.15.0/docs/oss/attribution.md)

docs/releases/v0.16.0.md

@@ -0,0 +1,86 @@
+# Promptkit v0.16.0
+
+This release adds explicit smoke diagnostics, configuration collections,
+namespace-scoped engine views, and expanded resource discovery. It also makes
+intentional breaking changes to configuration ownership and prompt selection.
+The minimum supported Go version remains 1.26.0.
+
+## Configuration Collections And Discovery
+
+Applications can compose caller-selected collections of backends, profiles,
+and prompts. Collection resources have qualified identities such as
+`support/answer`; root resources retain unqualified identities. `CollectionInput`
+provides a convenience for mounting a selected ConfigKit input into a collection.
+Promptkit now uses ConfigKit v0.8.0, including its optional source specification
+header support.
+
+`Engine.WithNamespaces` creates an immutable view with separate prompt, profile,
+and backend scopes for short caller-supplied IDs. Views share the underlying
+configuration, clients, capacity pools, and prepared-execution owner. Explicitly
+qualified IDs and configured references keep their exact identities.
+
+`ListProfiles` and `ListPrompts` support namespace filters. New `ListBackends`,
+`ListNamespaces`, and `ListCollections` methods expose captured resource and
+collection inventories. See the [collection format](../formats.md#source-headers-collections-and-resource-ids),
+[scoped-view guide](../consumers/pkg-promptkit.md#use-scoped-engine-views), and
+[discovery guide](../consumers/pkg-promptkit.md#discover-available-resources-and-collections).
+
+## Explicit Smoke Diagnostics
+
+`Engine.SmokeTest` provides backend reachability, authentication, text inference,
+and structured-output inference diagnostics. Reports provide factual
+observations, timing, and available usage rather than an overall readiness
+verdict. Optional Boolean facts distinguish negative observations from unknown
+or inapplicable checks; matching a JSON schema does not prove that a provider
+enforced it.
+
+A concluded diagnostic can report failure with a nil Go error. Invalid requests
+return ordinary errors, and caller cancellation can return partial observations
+with a recognizable context error. Diagnostics can contact providers and incur
+inference costs only when explicitly invoked; they are not automatic execution
+or release checks. See the [smoke diagnostic guide](../consumers/pkg-promptkit.md#run-explicit-smoke-diagnostics)
+and [API GoDoc](../../smoke_diagnostics.go).
+
+## Breaking Changes And Upgrade
+
+Consumers upgrading from v0.15.0 should follow the
+[configuration and API migration guide](../consumers/configuration-migration.md):
+
+- Supply backend and profile definitions explicitly. Promptkit no longer bundles
+  provider catalogs or the `promptkit-openrouter-standard` and
+  `promptkit-openrouter-auto` profiles. Conventional backend names do not install
+  definitions automatically. Applications choose and maintain their sources.
+- Convert version-keyed prompt mappings into one definition per prompt ID.
+  Prompt `version` is now an optional descriptive label, not a selector.
+  Choose deliberately which former definition to install.
+- Remove `PromptVersion` selectors from `RunRequest`,
+  `RequirementInspectionRequest`, and `SmokeTestRequest`, where applicable.
+  Change prompt inspection calls to `InspectPrompt(ctx, promptID)`. Update code
+  consuming version-bearing inspection, identity, result, or observation fields
+  to handle optional labels, including nil and explicitly empty labels.
+- When moving resources into collections, update stored backend, profile, and
+  diagnostic-default references to their intended qualified IDs. Mounting a
+  source does not rewrite references; profile inheritance remains within its
+  sibling mapping. Root configuration `version`, when supplied, is now a source
+  specification header, distinct from descriptive resource labels.
+
+The migration guide includes identity mappings and before/after prompt examples.
+The [maintained offline example](../../examples/go-library/prepare/main.go)
+demonstrates explicit composition, source-relative content, scoped preparation,
+and inventory without contacting a provider.
+
+After adapting configuration and API calls:
+
+```sh
+go get gitea.maximumdirect.net/eric/promptkit@v0.16.0
+```
+
+## Correctness And Efficiency
+
+Collection prompt materialization shares one lazy origin-metadata snapshot,
+avoiding repeated full metadata copies across collections. Internal prepared
+snapshots detach optional prompt labels. Smoke diagnostics preserve caller
+cancellation and report actual inference destinations and known-zero usage when
+no generation request was dispatched.
+
+[COPYRIGHT](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.16.0/COPYRIGHT) | [LICENSE](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.16.0/LICENSE) | [ATTRIBUTION](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.16.0/docs/oss/attribution.md)

docs/releases/v0.17.0.md

@@ -0,0 +1,68 @@
+# Promptkit v0.17.0
+
+This release adopts the vanity module path `go.maximumdirect.net/promptkit`
+and ConfigKit v1.0.0, standardizes configuration resources as identified lists,
+and removes descriptive prompt-version metadata. These are intentional breaking
+changes before v1. The minimum supported Go version increases from 1.26.0 to
+1.27.0 for the ConfigKit v1 migration.
+
+## Breaking Changes And Upgrade
+
+Consumers upgrading from v0.16.0 should follow the
+[configuration and API migration guide](../consumers/configuration-migration.md):
+
+- Change Promptkit imports from `gitea.maximumdirect.net/eric/promptkit` to
+  `go.maximumdirect.net/promptkit`. Update packages that exchange Promptkit
+  types together; old-path and vanity-path types are distinct.
+- Change ConfigKit imports to `go.maximumdirect.net/configkit` at v1.0.0 and
+  adopt its renamed public inspection types and fields where used.
+- Convert root and collection `prompts`, `profiles`, and `backends` from keyed
+  mappings to lists whose records contain an explicit `id`. Collections remain
+  keyed by namespace. Sibling inheritance now uses `extends: {id: parent}`;
+  later same-ID records replace lower records in full.
+- Change `ProfileSource` and `BackendSource` calls to pass slices of
+  `ProfileConfig` and `BackendConfig`, setting each record's `ID`.
+- Change source-root `configkit:` controls to `config:`. Imports are optional
+  unless marked `required: true`. Optional Promptkit specification headers now
+  use `version: 2` or `version: "2"`; headerless sources remain valid.
+- Remove descriptive prompt `version` values and code consuming `PromptVersion`,
+  `PromptIdentity.Version`, or corresponding `prompt_version` output members.
+  Collection `release_version` labels remain available. Prompt hashes may change
+  across this migration.
+
+After adapting Go imports, configuration, and API calls, use Go 1.27.0 or newer:
+
+```sh
+go get go.maximumdirect.net/promptkit@v0.17.0
+go mod tidy
+```
+
+Earlier tags retain their original module declarations. No permanent local
+replacement is needed for this cutover. See the
+[framework format reference](../formats.md) for current configuration semantics
+and the [maintained preparation example](../../examples/go-library/prepare/main.go)
+for explicit composition, source-relative imports, and scoped discovery.
+
+## Configuration Inspection And Execution
+
+Configuration inspection and diffs expose ConfigKit's composed values and
+provenance, including list-index paths. Promptkit applies domain normalization
+and execution defaults to owned interpretations without replacing the source
+origins needed for relative resources. Profile inspection and preparation expose
+their interpreted or effective settings. See the
+[migration guide](../consumers/configuration-migration.md#select-and-compose-the-configuration-you-need)
+for the distinction and inspection-path updates.
+
+## Maintainer Tooling
+
+Release tooling now uses the module-selected ReleaseKit v0.8.0 through `go tool`,
+with page-only Gitea delivery and saved-publication-attempt recovery. The
+[release procedure](../release.md) describes publication and recovery.
+
+The three maintained offline examples now have ordinary Go workflow tests,
+included in the standard suite. Their separate Python runner and custom release
+hook have been removed; Python is no longer a normal release prerequisite.
+The optional release-wiring diagnostic remains Python. See the
+[development guide](../development.md#maintainer-validation).
+
+[COPYRIGHT](https://forge.maximumdirect.net/base-layer/promptkit/src/tag/v0.17.0/COPYRIGHT) | [LICENSE](https://forge.maximumdirect.net/base-layer/promptkit/src/tag/v0.17.0/LICENSE) | [ATTRIBUTION](https://forge.maximumdirect.net/base-layer/promptkit/src/tag/v0.17.0/docs/oss/attribution.md)

docs/releases/v0.18.0.md

@@ -0,0 +1,65 @@
+# Promptkit v0.18.0
+
+Configuration inspection now exposes Promptkit's recorded field normalization
+and derived queue-capacity defaults through ConfigKit v1.1.0. Engines also
+include a default-on OpenRouter configuration. The minimum supported Go
+version remains 1.27.0.
+
+## Bundled OpenRouter Backend And Profiles
+
+New engines include an embedded OpenRouter backend and seventeen root profiles:
+ten latest-family aliases, the auto and free routers, and five auto-router
+cost-tier variants. Select a profile explicitly in a prompt or request; the
+bundle does not supply an implicit execution profile or contact providers
+during construction. The backend uses `OPENROUTER_API_KEY` when available under
+the existing optional-credential rules.
+
+The bundle is the lowest-priority underlay. Applications can replace its records
+or derive named variants through ordinary configuration, and external collections
+remain supported. Cost-tier variants inherit the effective `auto` profile, so
+replacing that parent changes their inherited target while retaining plugins.
+The [bundle reference](../formats.md#bundled-openrouter-configuration) owns the
+inventory and composition details.
+
+Explicit smoke diagnostics now have OpenRouter authentication metadata and an
+inference fallback profile. They remain opt-in and may incur provider costs.
+Latest aliases and routers can change upstream; configured slugs do not pin
+the concrete model, pricing, or availability. See the
+[consumer guide](../consumers/pkg-promptkit.md#construct-an-engine) for selection
+and opt-out examples.
+
+## Recorded Configuration Normalization
+
+Inspection shows normalized backend and profile settings, prompt metadata, and
+configured message roles and cache controls, with field-level resolution
+history. Existing source evidence remains available, and unchanged content-file
+origins continue to select their resource context.
+
+Configuration diffs compare normalized effective values, so spelling-only
+changes that normalize identically may no longer appear. Lists remain atomic in
+diffs. Inspection does not include backend-to-profile completion or request
+overrides; see the [configuration inspection guide](../consumers/pkg-promptkit.md#inspect-the-captured-configuration).
+
+## Compatibility And Upgrade
+
+New engines install a bundled OpenRouter backend and root profiles by default.
+Inventory listings and inspection sources therefore change, and references to
+bundled IDs that previously failed may now resolve. Pass
+`promptkit.WithBundledOpenRouter(false)` to retain application-only behavior.
+Caller records can replace bundled definitions through ordinary composition;
+same-ID replacements replace complete records rather than merging selected fields.
+No configuration-format migration is required from v0.17.0.
+Consumers that display or compare configuration inspection results should account
+for normalized values and derived defaults replacing their earlier authored
+representations. Request-time overrides and caller-appended messages retain
+their existing processing.
+
+```sh
+go get go.maximumdirect.net/promptkit@v0.18.0
+```
+
+See the [framework format reference](../formats.md) for current configuration
+semantics. Consumers upgrading from versions before v0.17.0 should also follow
+the [configuration and API migration guide](../consumers/configuration-migration.md).
+
+[COPYRIGHT](https://forge.maximumdirect.net/base-layer/promptkit/src/tag/v0.18.0/COPYRIGHT) | [LICENSE](https://forge.maximumdirect.net/base-layer/promptkit/src/tag/v0.18.0/LICENSE) | [ATTRIBUTION](https://forge.maximumdirect.net/base-layer/promptkit/src/tag/v0.18.0/docs/oss/attribution.md)

docs/releases/v0.7.0.md

@@ -63,8 +63,8 @@ Every engine now includes the reserved `rakestrawhome` backend, identified by
 selects that backend. Consumers can use the maintained endpoint, credential,
 capacity, and model defaults without registering either definition themselves.
 
-See the [built-in backend and profile catalogs](../formats.md#built-in-backends)
-and the [consumer adoption example](../consumers/pkg-promptkit.md#use-the-rakestrawhome-built-in-profile)
+See the [backend format reference](../formats.md#backend-definitions)
+and the [current identity migration mapping](../consumers/configuration-migration.md#identity-mapping)
 for the current contracts.
 
 ## Structured Generation Errors
@@ -126,8 +126,8 @@ cycle-safe, and resolved afresh for ordinary operations. Prepared execution
 freezes the fully resolved target. The selected leaf ID remains public while
 effective execution settings reflect the resolved chain.
 
-See the [profile inheritance format reference](../formats.md#profile-inheritance),
-the [consumer alias example](../consumers/pkg-promptkit.md#alias-a-built-in-profile),
+See the [profile inheritance format reference](../formats.md#resource-inheritance),
+the [current identity migration mapping](../consumers/configuration-migration.md#identity-mapping),
 and the [`Profile` GoDoc](../../types.go) for exact merge and validation
 behavior.
 

docs/releases/v0.8.0.md

@@ -0,0 +1,109 @@
+# Promptkit v0.8.0
+
+This supplemental changelog and migration guide summarizes the consumer-facing
+changes from `v0.7.0` to `v0.8.0`. The annotated `v0.8.0` tag is the
+authoritative release record. Exact current contracts belong to the linked
+GoDoc and durable documentation.
+
+## Summary
+
+`v0.8.0` activates Promptkit's bounded output-repair workflow:
+
+- failed nonempty-text, JSON, and JSON Schema validation can make a limited
+  number of corrective model calls;
+- corrective calls preserve the original rendered conversation, effective
+  target, session, structured-output contract, and backend capacity policy;
+- results report cumulative usage and the number of corrective calls actually
+  made; and
+- explicitly empty OpenAI-compatible response content now reaches output
+  validation instead of being classified as a malformed provider envelope.
+
+## Compatibility
+
+This release adds no public declarations or fields and removes none. Existing
+source code remains source-compatible.
+
+The behavior of the existing `OutputContract.RepairAttempts` field and prompt
+YAML `repair_attempts` field has changed. A positive value now authorizes real
+additional model calls after eligible validation failures; earlier releases
+accepted the field but the public engine remained single-pass. Consumers that
+set a positive value should expect additional latency, token usage, and
+provider cost when repair is needed.
+
+Repair budgets must now be between zero and three. A positive budget requires
+`basic`, `json`, or `json_schema` validation. Values above three and a positive
+budget paired with `none` are invalid contracts rather than ignored settings.
+
+An explicitly present empty or whitespace-only string returned by the built-in
+OpenAI-compatible client is now a completed generation candidate. `none`
+validation permits it, while `basic`, `json`, and `json_schema` classify it
+under their ordinary validation rules and may repair it when configured.
+Missing, `null`, or non-string content remains a malformed provider response.
+
+## Upgrade
+
+Update the module dependency with:
+
+```sh
+go get gitea.maximumdirect.net/eric/promptkit@v0.8.0
+go mod tidy
+```
+
+Review every prompt definition and request override that sets a positive repair
+budget. Use zero or omit the field to retain single-pass execution. Ensure each
+positive budget is no greater than three and uses an eligible validation mode,
+then run the consuming project's ordinary and race-enabled tests.
+
+## Bounded Output Repair
+
+`repair_attempts` counts corrective calls in addition to the initial model
+call. Promptkit validates each completed candidate, stops at the first valid
+one, and never exceeds the configured bound. If every candidate remains
+invalid, the run completes successfully with the final candidate and its
+failed validation result rather than returning an operational error.
+
+Each correction starts from the original rendered messages and includes only
+the latest invalid candidate and latest validation diagnostics. JSON Schema
+mode retains the provider-native structured-output request as its first line of
+defense. Promptkit performs only deterministic structural validation; a valid
+response is not necessarily factual or correct for an application's domain.
+
+Usage in the final result is cumulative across the initial response and every
+completed corrective response. `ValidationResult.RepairAttempts` reports the
+number of corrective calls actually made. Corrective generation failures use
+the same public generation-error categories and structured provider details as
+an initial generation failure.
+
+See the [output-contract format reference](../formats.md#output-contract), the
+[consumer repair example](../consumers/pkg-promptkit.md#repair-a-structured-result),
+and the [`OutputContract` and `ValidationResult` GoDoc](../../types.go) for the
+current contracts.
+
+## Explicit Empty Content
+
+The built-in OpenAI-compatible client now distinguishes an explicitly present
+empty string from a missing or malformed `content` field. This aligns built-in
+and injected clients by letting the selected output contract decide whether an
+empty candidate is acceptable, invalid, or eligible for repair.
+
+See the
+[OpenAI-compatible response contract](../integrations/openai-compatible-chat.md#response-handling)
+for the exact envelope behavior.
+
+## Public API Changes
+
+None. This release activates and tightens the documented behavior of existing
+fields.
+
+## Consumer Action
+
+- Remove or set `repair_attempts` to zero where execution must remain
+  single-pass.
+- Keep every positive repair budget at three or fewer and pair it with
+  `basic`, `json`, or `json_schema` validation.
+- Account for additional latency, usage, and provider cost when enabling
+  repair.
+- Continue checking the returned validation status because bounded repair can
+  exhaust without producing a valid candidate.
+- Review workflows that previously treated explicit empty provider content as
+  a generation error.

docs/releases/v0.9.0.md

@@ -0,0 +1,124 @@
+# Promptkit v0.9.0
+
+This supplemental changelog and migration guide summarizes the consumer-facing
+changes from `v0.8.0` to `v0.9.0`. The annotated `v0.9.0` tag is the
+authoritative release record. Exact current contracts belong to the linked
+GoDoc and durable documentation.
+
+## Summary
+
+`v0.9.0` adds stateless request-message composition and separates maintained
+provider data from Promptkit's core implementation:
+
+- callers can append already-rendered messages to a configured prompt for
+  application-owned conversations and semantic correction workflows;
+- the public package now publishes constants for the four supported text-chat
+  roles, and prompt definitions use the same normalized role vocabulary; and
+- the OpenRouter and Rakestrawhome backend/profile catalogs now come from two
+  independently versioned Go module dependencies.
+
+## Compatibility
+
+This release adds one field and four constants to the public API and removes no
+public declarations. Existing keyed `RunRequest` literals that omit
+`AppendedMessages` retain their behavior. Adding the field changes the struct
+shape, so consumers using positional `RunRequest` literals must convert them
+to keyed literals.
+
+Message roles in prompt definitions are now trimmed, lowercased, and required
+to be `developer`, `system`, `user`, or `assistant`. Definitions using another
+role that earlier releases accepted as an arbitrary nonblank string now fail
+prompt loading. In particular, the text-only message contract does not support
+`tool` or the deprecated `function` role. Otherwise valid roles with different
+case or surrounding whitespace are normalized rather than rejected.
+
+The catalog extraction preserves Promptkit's public API, built-in backend and
+profile IDs, configuration, precedence, credential handling, and capacity
+behavior. Consumers do not import or register either catalog themselves, and
+no configuration migration is required. Promptkit now selects two
+independently versioned data dependencies and guarantees only the catalog
+versions selected and tested by this Promptkit release.
+
+## Upgrade
+
+Update the module dependency with:
+
+```sh
+go get gitea.maximumdirect.net/eric/promptkit@v0.9.0
+go mod tidy
+```
+
+Convert any positional `RunRequest` literals to keyed literals. Review prompt
+definitions for unsupported roles, then run the consuming project's ordinary
+and race-enabled tests.
+
+## Appended Request Messages
+
+`RunRequest.AppendedMessages` accepts caller-owned `RenderedMessage` values
+that Promptkit validates, copies, and appends after every rendered
+prompt-definition message in caller order. Promptkit does not template this
+content, retain conversation state between calls, impose a retry policy, or
+apply a message-count, byte-size, token, or context-window limit. Upstream
+rejections continue through the ordinary generation-error boundary.
+
+This primitive supports application-owned conversation continuations and
+domain-aware correction loops while preserving Promptkit's existing
+preparation, hashing, prepared-execution, structural repair, backend-capacity,
+credential, and cancellation behavior. Appended content can include sensitive
+model output or application feedback; prepared values expose the complete
+effective messages by design, while default request formatting reports only
+the appended-message count.
+
+See the
+[consumer appended-message example](../consumers/pkg-promptkit.md#append-already-rendered-messages),
+the [`RunRequest`, `RenderedMessage`, and `CacheControl` GoDoc](../../types.go),
+and the [OpenAI-compatible request contract](../integrations/openai-compatible-chat.md#request-body)
+for current behavior.
+
+## Supported Message Roles
+
+The new `RoleDeveloper`, `RoleSystem`, `RoleUser`, and `RoleAssistant`
+constants identify the complete role vocabulary accepted by Promptkit's
+text-chat message model. The same validation and normalization now apply to
+prompt-definition messages and request-supplied appended messages. Promptkit
+does not translate between roles; provider- or model-specific rejection of an
+otherwise supported role remains an upstream generation error.
+
+See the [message format reference](../formats.md#messages-and-templates) for
+the canonical prompt-definition contract.
+
+## Independently Versioned Backend Catalogs
+
+Promptkit imports immutable catalog data for its maintained OpenRouter and
+Rakestrawhome backends and profiles. Engine construction validates and
+assembles both catalogs behind the existing built-in registry and profile
+precedence rules. Promptkit no longer keeps duplicate embedded profile assets
+or hard-coded definitions for those maintained backends.
+
+The module versions in Promptkit's `go.mod` identify the catalog releases
+tested with this release. The [backend format reference](../formats.md#backend-definitions)
+remains the canonical consumer contract, while the [internal source documentation](../internal/sources.md#profiles)
+describes the dependency boundary.
+
+## Public API Changes
+
+The release adds:
+
+- `RunRequest.AppendedMessages`;
+- `RoleDeveloper`;
+- `RoleSystem`;
+- `RoleUser`; and
+- `RoleAssistant`.
+
+No public declaration was removed.
+
+## Consumer Action
+
+- Convert positional `RunRequest` literals to keyed literals.
+- Replace unsupported prompt-definition roles with an appropriate supported
+  role, or keep richer tool-call protocols in an application-owned client.
+- Treat appended messages and prepared effective messages according to the
+  application's sensitive-data policy.
+- Do not add direct catalog imports or registration calls; existing Promptkit
+  construction and configuration remain correct.
+- Run consumer ordinary and race-enabled tests after updating the module.

docs/roadmap/audit/01-correctness.md

@@ -0,0 +1,131 @@
+# Functional Correctness Audit
+
+Reviewed 2026-11-04 against commit `d9f6aad`.
+
+## Scope And Evidence
+
+Reviewed the public facade and conversions; ConfigKit handoff, normalization,
+presence-aware target assembly, collections and namespace selection; artifact
+loading and template rendering/requirement discovery; ordinary and single-use
+prepared execution; output validation and repair; backend admission and active
+permits; HTTP request/response handling and transient retries; execution
+observations; and smoke-test selection and reporting. Contracts were checked
+against GoDoc, the format and integration references, internal documentation,
+and relevant existing tests. Source loading before the ConfigKit handoff remains
+ConfigKit's responsibility.
+
+Validation:
+
+- `go test -race -vet=off ./...` passed, including the maintained examples.
+- `go vet ./...` passed.
+- An offline, temporary public-API reproduction outside the repository confirmed
+  the number conversion and captured schema payload described below. It used
+  `ProfileSource`, `Prepare`, a runtime override, and `WithSchemaFS` with
+  `fstest.MapFS`; no provider was contacted.
+
+No production code or tests were changed. This audit does not establish provider
+availability or acceptance of arbitrary JSON Schema features.
+
+## 1. Provider Source Helpers Silently Change Exact JSON Numbers
+
+**Priority:** P2. Confirmed data-conversion defect.
+
+**Code:** [providerValueForMapping](../../../provider_values.go:51), used by
+[ProfileSource](../../../profiles.go:33) and
+[BackendSource](../../../backends.go:100).
+
+**Circumstances:** A caller supplies a valid `json.Number` in provider
+`ExtraParams` that cannot be represented exactly by the helper's chosen native
+number type. After signed and unsigned integer parsing fail, the helper converts
+the value to `float64` without checking whether that conversion loses precision.
+The same problem affects sufficiently precise fractional values.
+
+**Expected:** A source helper preserves the meaning of a supported provider
+value, or reports that the value cannot be represented. It must not silently
+substitute a different number. This concerns PromptKit's own convenience-source
+conversion, not additional validation of ConfigKit's raw sources.
+
+**Actual / reproduction:** With
+`ExtraParams: map[string]any{"number": json.Number("18446744073709551617")}`
+on a profile, `NewEngine` succeeds. JSON encoding of
+`Prepare(...).EffectiveModelParams.ExtraParams` produces:
+
+```json
+{"number":18446744073709552000}
+```
+
+Supplying the identical value through `RunRequest.Execution.ExtraParams`
+instead produces:
+
+```json
+{"number":18446744073709551617}
+```
+
+The built-in client serializes those effective values directly into the request,
+so configuration-source selection changes the actual provider parameter.
+`BackendSource` shares the same converter. No high-precision numeric preservation
+claim needs to be added for arbitrary source formats to correct this discrepancy.
+
+**Potential remediation:** Preserve exact numbers through a numeric
+representation supported by ConfigKit, if available. If that contract cannot
+represent a particular `json.Number` exactly, surface a deferred source error
+rather than returning a rounded float or a numeric-looking string. Check numeric
+value equivalence rather than textual spelling: alternate exponent spellings
+need not be preserved. Add focused regression coverage for both source helpers
+and contrast with the already-preserving runtime path.
+
+## 2. External Schema Resources Do Not Reach The Provider Constraint
+
+**Priority:** P2. Confirmed schema-packaging gap; remediation requires choosing
+how provider-facing schemas accommodate the existing local-reference feature.
+
+**Code:** [preparedValidation.SchemaDocument](../../../internal/validate/standard_validator.go:63),
+[StandardValidator.PrepareValidation](../../../internal/validate/standard_validator.go:74),
+[FSValidator.PrepareValidation](../../../internal/validate/standard_validator.go:126),
+and [structuredOutputFromValidationPlan](../../../internal/usecase/runner.go:419).
+
+**Circumstances:** A JSON Schema output contract uses a root schema referring to
+another file within the configured schema source. The
+[schema format contract](../../formats.md#schemas) supports such references.
+Preparation compiles the transitive resources locally, but `SchemaDocument`
+exposes only the unchanged root document. The runner always uses that document
+for the native provider constraint in JSON Schema mode.
+
+**Expected:** The provider-facing constraint should carry the schema resources
+needed to interpret the supported local references, or there should be an
+explicitly chosen way to retain local validation without submitting an
+unresolvable native constraint. This is not a promise that every provider accepts
+every schema dialect feature.
+
+**Actual / reproduction:** With these files under `WithSchemaFS`:
+
+```text
+root.json:  {"type":"object","properties":{"x":{"$ref":"child.json"}},"required":["x"]}
+child.json: {"type":"integer"}
+```
+
+`Prepare` succeeds, but `PreparedRun.StructuredOutput.JSONSchema.Schema` contains
+only the root object, including the literal `$ref: "child.json"`. No child
+document, bundled definition, or reference rewrite is present. The HTTP adapter
+passes that object unchanged to `response_format.json_schema.schema`. An upstream
+provider has no access to the application filesystem or private `fs.FS`, so it
+cannot resolve that reference from the request. A provider requiring resolvable
+native schemas can reject generation before local validation or repair is
+reached. No live-provider rejection was tested.
+
+The internal source document accurately describes the existing root-only
+projection; the gap is between the supported local schema-resource graph and
+its mandatory projection into an outbound constraint, rather than a failure of
+local compilation or prepared-resource freezing. Self-contained schemas are
+unaffected.
+
+**Potential remediation:** Produce a self-contained provider schema from the
+operation's captured resources, preserving reference semantics, identifiers,
+anchors, and recursion. Avoid naive recursive textual substitution. An
+alternative is an explicit choice to use local-only schema validation when
+native submission cannot represent the resource graph. Any alternative that
+restricts existing schema sources or silently disables native structured output
+requires a deliberate compatibility/policy decision. Protect the selected
+behavior with an offline test combining transitive schema preparation and the
+actual outgoing request, while retaining frozen local validation for
+`RunPrepared` and repair.

docs/roadmap/audit/02-efficiency.md

@@ -0,0 +1,147 @@
+# Algorithmic Efficiency Audit
+
+Reviewed 2026-11-04 against commit `0d27252`.
+
+## Scope And Method
+
+Reviewed PromptKit-owned configuration assembly, provider-value conversion and
+copying, rendering and requirement analysis, inventory selection, prepared-run
+ownership, validation orchestration, repair formatting, retry coordination,
+observations, and capacity management. Findings below follow from implementation
+and operation counts; no benchmarks or live-provider calls were needed.
+
+The public facade delegates to internal components. ConfigKit supplies the final
+configuration; PromptKit constructs immutable runtime indexes and captured
+templates. Execution uses operation-local validation plans, shared backend
+capacity, and a generation coordinator. Inspection reads the captured snapshot.
+ConfigKit and ReleaseKit internals are outside this repository audit.
+
+### Deliberate Choices Preserved
+
+Checked current internal documentation and relevant retired plans in Git history,
+including the performance plan at `8a5c0de`, configuration adoption at `5f7b194`,
+and collections at `89af2a1`. This audit does not reopen:
+
+- lexical inventory ordering, recursive provider-map overlay semantics, or
+  conservative requirement-discovery semantics;
+- captured parsing, per-render artifact-text reuse, operation-local schema
+  compilation, frozen retry payloads, and linked-list FIFO permits with
+  constant-time cancellation removal;
+- the existing JSON depth/work bounds, output/diagnostic limits, or copies
+  deliberately retained to protect public and injected-collaborator ownership;
+- synchronous filesystem/template/schema operations or the selected retry and
+  repair budgets.
+
+The opportunities below concern implementation work that those choices do not
+require. Complexity statements use expected constant-time Go map operations and
+treat identifier comparison/hashing cost separately. No new public performance
+guarantee, cache, caller limit, or release gate is proposed.
+
+## 1. Condition Evidence Unions Can Require Quadratic Work
+
+**Priority:** P2. Requirement-inspection path; most relevant to deeply nested
+conditions rather than ordinary small templates.
+
+**Code:** [evaluateCommand](../../../internal/prompt/requirements.go:142) and
+[mergeConditionEvaluations](../../../internal/prompt/requirements.go:318).
+
+Each comparison recursively evaluates its operands, then allocates fresh
+`missingVariables` and `contentInputs` maps and copies both operands' accumulated
+evidence. A left-heavy expression with distinct missing caller keys, such as
+successively nesting `eq (eq .v0 .v1) .v2`, carries an increasingly large missing
+key set through each ancestor. Template parsing accepts these expressions;
+inspection can encounter them even when their eventual execution would fail or
+the inspection result is partial.
+
+For an expression of size n, successive unions copy approximately
+2 + 3 + ... + n entries: **Theta(n²) total set work and cumulative allocation**.
+The final unique evidence has only O(n) entries. This is separate from the
+previously fixed repeated syntax-subtree scan: visiting each syntax node once
+does not eliminate repeated copies of the evidence produced by those nodes.
+
+**Opportunity:** Transfer ownership of temporary sets and merge the smaller set
+into the larger one, rather than copying both at every comparison. For this
+expression family that is O(n); small-to-large merging gives O(n log n) aggregate
+set-insertion work across general expression trees. Another option is an
+operation-local evidence accumulator that records each discovery directly,
+while keeping only condition truth/limitation state on recursive return values.
+Choose the simpler approach after checking temporary-set ownership. Preserve
+current grammar, classifications, deduplication, sorted output, cancellation,
+and partial-result behavior; interpreting more expressions is not necessary.
+
+## 2. Recursive Provider Overlay Repeatedly Clones Descendant Subtrees
+
+**Priority:** P3. Construction-time improvement with an existing depth bound.
+
+**Code:** [overlayProviderMap](../../../internal/configuration/targets.go:131).
+
+The helper deep-copies the entire lower map before considering overrides. When
+both values at a key are maps, it recursively invokes itself on a descendant
+that was already copied, and copies that whole descendant again. The first copy
+is then replaced and discarded. All descendants on an overlapping path can be
+copied once per ancestor.
+
+Let N be the lower tree's node count and D its overlapping map depth. Node-copy
+work can reach **O(ND)**. For a narrow chain it is the triangular sum
+D + (D-1) + ... + 1. A wide K-entry leaf below D overlapping ancestors is cloned
+roughly D times instead of once. `jsonvalue.CopyMap` also sorts map keys, so the
+same wide leaf's sorting work is repeated at those ancestors.
+
+The existing maximum container depth of 100 is important: this is **not an
+unbounded quadratic-in-N claim under today's fixed limits**. It is an avoidable
+depth-dependent multiplier in traversal and cumulative allocations within those
+limits, multiplied again when multiple profiles overlay the same backend. That
+bound and the recursive-map/value-replacement semantics remain selected choices.
+
+**Opportunity:** Clone the lower tree once, then overlay into the owned clone
+without re-cloning existing descendant maps. Copy incoming replacement values
+where needed to preserve isolation from configuration inputs. Alternatively,
+construct the merged output in one traversal. Either removes the repeated
+ancestor copies: expected O(N + M) merge traversal for lower/upper trees of sizes
+N and M, apart from the retained validation/sorting costs. Preserve null and
+array replacement, empty-map behavior, input ownership, and existing JSON
+validation/bounds; do not replace ConfigKit's configuration machinery.
+
+## 3. Filtered Listings Copy And Scan The Complete Inventory
+
+**Priority:** P3. Scalability opportunity for engines with many namespaces.
+
+**Code:** [filterIdentityList](../../../internal/engine/engine.go:306),
+[ListPrompts](../../../internal/engine/engine.go:270), and the
+[configuration inventory methods](../../../internal/configuration/listing.go:14).
+
+A named-namespace request first obtains a detached copy of all N identities and
+then scans it to build its k matching results. Even an absent namespace costs
+**Theta(N) time and O(N) temporary storage**, rather than work proportional to
+the selected inventory. Listing C namespaces separately costs Theta(CN); when
+each has a bounded number of entries, that becomes quadratic in total inventory
+size despite only O(N) total returned entries.
+
+The collection implementation plan at `89af2a1` expressly allowed namespace
+buckets or sorted ranges to avoid repeated full scans; it did not require the
+present whole-inventory-copy algorithm. The lexical ordering and caller-owned
+result contracts do not prevent indexed selection.
+
+**Opportunity:** Capture per-kind namespace buckets or ranges alongside existing
+construction-time inventories, then copy only the selected entries. Buckets
+allow expected O(1 + k) selection/copy; binary search over sorted full IDs can
+support named-prefix ranges in O(log N + k). Keep unqualified and all-ID modes
+correct without assuming unqualified entries form one contiguous lexical range.
+Resolve filters before requesting the inventory, and retain exact namespace
+matching, full IDs, ordering, context behavior, and independent returned slices.
+An all-ID request must still spend O(N) to return N owned entries; that is not a
+defect. No public pagination API or mutable cache is needed.
+
+## Review Outcome
+
+These three opportunities are supported by direct operation-count analysis;
+they are not measured latency claims. The largest asymptotic issue is growing
+condition-evidence copying. The overlay finding is explicitly bounded, and the
+listing finding matters as independent namespace count grows.
+
+No additional ordinary correctness defect was established in this pass. The
+findings in [the correctness audit](01-correctness.md) remain separate and were
+not duplicated here. No production code, tests, dependencies, or established
+behavior were changed, so the successful race-suite/vet evidence recorded there
+still applies to the unchanged implementation; those checks were not rerun just
+for this documentation change.

docs/roadmap/audit/03-complexity.md

@@ -0,0 +1,216 @@
+# Complexity And Maintainability Audit
+
+Reviewed 2026-11-04 against commit `5adc1c1`.
+
+## Method And Architectural Assessment
+
+Used the local codebase-memory graph's cyclomatic/cognitive complexity,
+function-length, and call-relationship tools to locate candidates. Reviewed
+functions and methods, including test hotspots, rather than relying on
+function-only rankings. Inspected implementation, callers, owning interfaces,
+and current architecture/internal documentation before selecting findings.
+Metrics are analyzer-specific indicators, not correctness proofs, acceptance
+thresholds, or requirements to split every large switch.
+
+PromptKit's overall boundaries remain useful: a public facade adapts values and
+errors; configuration interprets ConfigKit's final handoff and captures runtime
+indexes; the runner coordinates preparation and execution; generation owns
+attempts/retries; capacity owns admission and permits; transport and schema
+adapters own external protocols. No repository-wide architectural rewrite or new
+public abstraction is warranted by this review.
+
+Representative hotspot metrics from the current graph:
+
+| Function | Cyclomatic | Cognitive |
+| --- | ---: | ---: |
+| `requirementWalker.visit` | 50 | 137 |
+| `validateEffective` | 39 | 132 |
+| `requirementWalker.evaluateCommand` | 26 | 103 |
+| `providerValueForMapping` | 33 | 84 |
+| `jsonvalue.copyValue` | 29 | 66 |
+| `loaded.buildExecutionProfiles` | 27 | 58 |
+| `loaded.materializePrompt` | 30 | 52 |
+| `Coordinator.generate` | 24 | 49 |
+
+The findings below are behavior-preserving refactoring opportunities, not
+confirmed bugs or automatic release blockers. They cover both package-local
+complexity and interactions between packages. No production code, tests, or
+dependencies were changed, and no empirical testing or benchmarks were needed.
+
+## 1. Consolidate Schema Compilation And Express The Runner's Actual Interface
+
+**Code:** [validator interfaces](../../../internal/validate/validator.go:10),
+[schema implementations](../../../internal/validate/standard_validator.go:45),
+[runner preparation](../../../internal/usecase/runner.go:398), and
+[engine assembly](../../../internal/engine/engine.go:133).
+
+**Complexity:** The validator maintains direct validation and prepared
+validation, with separate filesystem and `fs.FS` compilation paths. Compiler
+construction, resource registration, error wrapping, and cancellation checks
+are repeated across direct schema-validation methods and `PrepareValidation`.
+The package's constructors return `Validator`, although the runner actually
+requires `ValidationPreparer` and discovers it through a runtime type assertion.
+An implementation satisfying the declared constructor dependency can therefore
+still be unsuitable for the execution path.
+
+**Opportunity:** Make the runner's preparation dependency explicit in its
+internal field/constructor types, updating assembly and focused test
+collaborators together. Retain the direct validation interface where callers
+actually need it. Consolidate compilation behind one internal operation, with
+small source-specific loading/URL/loader adapters; direct and prepared callers
+can reuse that operation rather than separately implementing registration and
+compilation. No public validator API or generic source framework is needed.
+
+**Preserve:** Direct validation currently rejects malformed JSON before trying
+to compile its schema; simply implementing direct `Validate` as eager
+`PrepareValidation` followed by plan validation would change that error/I/O
+ordering. Keep that short-circuit, operation-local compiled resources, prepared
+freezing, schema dialect and containment rules, synchronous cancellation
+precedence, and the runner's existing nil-validator behavior where used.
+Coordinate schema representation changes with correctness-audit finding 2,
+without treating this refactor alone as its fix.
+
+## 2. Decompose Configuration Validation And Resource Materialization
+
+**Code:** [validateEffective](../../../internal/configuration/load.go:410),
+[materializePrompt](../../../internal/configuration/prompts.go:58),
+[buildExecutionProfiles](../../../internal/configuration/targets.go:14), and
+[resolution callbacks](../../../internal/configuration/resolve_prompts.go:10).
+
+**Complexity:** `validateEffective` combines namespace checks, reference indexes,
+backend settings, capacity arithmetic, smoke metadata, profile bounds, and prompt
+references in one nested traversal. Prompt materialization then combines input
+validation, presence projection, content-file I/O, message validation, template
+compilation, output-contract construction, hashing, and inspection assembly.
+Target assembly is another long presence-state decision chain. Reviewing a
+single field change requires following several distinct responsibilities inside
+large functions.
+
+**Opportunity:** Keep the top-level traversal/index preparation visible, but
+delegate coherent record-level work to backend, profile, and prompt helpers
+with explicit record/path/index arguments. Isolate prompt input projection,
+message materialization/compilation, and output-contract construction into
+small functions returning the values the caller needs. Keep target presence
+interpretation in one identifiable helper rather than distributing it between
+new callers. Organize files by those responsibilities only where that helps
+navigation; moving unchanged code between files is not itself the improvement.
+
+**Preserve:** Validate only final ConfigKit-provided values. Keep recorded
+normalization, field presence/null distinctions, exact references, error
+identities, source-relative resource contexts, lazy metadata capture, and
+construction-time template/hash capture. Materialization legitimately combines
+validation with resource access; this finding does not require moving every
+check into a new validation pass or aggregating formerly fail-fast errors.
+Avoid a reflection-driven field schema that obscures the domain rules.
+
+## 3. Untangle Requirement Traversal, Expression Evaluation, And Evidence
+
+**Code:** [requirement walker](../../../internal/prompt/requirements.go:356),
+[syntax dispatch](../../../internal/prompt/requirements.go:451),
+[condition evaluation](../../../internal/prompt/requirements.go:93), and
+[command traversal](../../../internal/prompt/requirements.go:652).
+
+**Complexity:** The approximately thousand-line file contains mutually recursive
+syntax walking and partial expression evaluation, branch reachability, caller
+scope tracking, template-call handling, evidence promotion, issue construction,
+and result sorting. `if` and `with` repeat branch orchestration. `input`, `index`,
+root variables, and skipped-path behavior appear in both evaluation and
+conservative traversal. Correctness depends on the interaction of return-value
+flags and mutable walker state such as `rootKnown`, `dependencies`, and the
+unknown-effect flags, not just the local switch case.
+
+**Opportunity:** Separate the existing roles into small internal units: syntax
+recognition/evaluation, branch/scope traversal, and evidence/result accumulation.
+Factor shared `if`/`with` orchestration while making their different body-dot
+rules explicit. Reuse recognition of supported command shapes instead of
+duplicating it between evaluative and inventory-only paths. Make existing
+template scope entry/exit state handling explicit. Keep a small enum/value model
+rather than adding a general expression interpreter or pluggable AST framework.
+
+**Preserve:** The deliberately conservative recognition scope, skipped-path
+inventory, optional `index` behavior, named-template limitations, incomplete
+results, and cancellation semantics. Traversal and evaluation are not
+interchangeable: inactive paths must retain recognizable names without
+evaluating caller-derived names. Any shared helper must express that distinction.
+The evidence-union efficiency finding in [audit 02](02-efficiency.md) is related
+but separate; a mechanical file split would not fix it.
+
+## 4. Clarify Ownership And Policies Across The JSON Value Walkers
+
+**Code:** [source mapping conversion](../../../provider_values.go:31),
+[bounded JSON copying](../../../internal/jsonvalue/jsonvalue.go:67), and
+[public-value copying](../../../convert.go:406).
+
+**Complexity:** Three recursive reflection-based paths independently handle
+interfaces/pointers, maps, sequences, scalars, and nil values. Each adds its own
+policy: source conversion translates into ConfigKit-compatible containers and
+defers unsupported-value errors; `jsonvalue` validates and bounds owned copies;
+facade copying detaches already-valid values while preserving their types.
+Similar-looking switches conceal different assumptions and failure behavior.
+The numeric conversion defect in correctness-audit finding 1 demonstrates a
+concrete risk at one of these boundaries, but is not a new finding here.
+
+**Opportunity:** First make the three responsibilities explicit at their call
+boundaries. Isolate source-number encoding from structural traversal, and
+identify reusable scalar/indirection/container-copy primitives only where their
+semantics truly match. Prefer removing a redundant traversal path or reusing an
+existing trusted-value copy operation over maintaining another policy-rich
+reflection switch. Keep ConfigKit-specific mapping encoding in its adapter;
+do not teach the source-neutral domain about ConfigKit.
+
+**Preserve:** Supported concrete types, nil behavior, cycle/depth/work handling,
+caller ownership, and deferred source errors. In particular, passing source
+builders through rejecting `CopyMap` would introduce validation before the
+ConfigKit handoff and is not an acceptable consolidation. Nor should an
+infallible facade projection acquire new rejection behavior merely to reuse a
+validator. A universal callback-configured reflection framework could increase
+complexity; retain separate small adapters where their policies differ.
+
+## 5. Extract Attempt Accounting From Generation Control Flow
+
+**Code:** [Coordinator.generate](../../../internal/generation/coordinator.go:98),
+[generateInjected](../../../internal/generation/coordinator.go:203),
+[execute](../../../internal/generation/coordinator.go:303), and
+[diagnostic evidence consumers](../../../internal/usecase/smoke_tester.go:246).
+
+**Complexity:** The coordinator interleaves permit acquisition, dispatch,
+cancellation, retry decisions, observer events, diagnostic evidence, and
+presence-aware usage accumulation. The built-in and injected paths repeat
+generation-number setup, evidence finishing, attempt projection, and usage
+projection. The injected path also repeats its non-dispatched cancellation
+reporting. `execute` returns five parallel values, increasing the amount of
+state the caller must keep aligned while constructing two reporting forms.
+
+**Opportunity:** Return one small internal attempt-result value containing
+outcome and timing/dispatch facts, and extract common projection/accounting
+operations where evidence is genuinely shared. Keep the retry loop primarily
+about eligibility, budget, waits, and terminal outcomes. Keep injected and
+built-in dispatch strategies explicit rather than manufacturing HTTP facts for
+injected clients or introducing an extensible retry framework.
+
+**Preserve:** Release active permits before callbacks/backoff, one-shot injected
+dispatch, frozen built-in requests, usage-presence distinctions, private
+diagnostic evidence, ordinary observer boundaries, and retention of decoded
+facts before cancellation discards ordinary results. The HTTP adapter should
+continue owning transport facts; accounting extraction must not push provider
+protocol knowledge into the runner or observation package.
+
+## Prioritization And Non-Findings
+
+Start with the schema interface/compilation consolidation and configuration
+helpers: their responsibilities provide comparatively clear refactoring seams.
+The walker and generation coordinator need narrower, behavior-preserving units
+because their state transitions are more subtle. Coordinate source conversion
+cleanup with the independently required number-preservation fix.
+
+High scores alone did not justify findings for explicit public error-category
+switches, presence-aware source-field encoding, regular-file/schema read loops,
+or table-driven tests. Those branches often directly express selected contracts.
+Large facade files also include substantial GoDoc and straightforward value
+projection, which is not evidence that public/internal separation has failed.
+Keep useful test scenarios; do not rewrite tests solely to reduce their metrics.
+
+No additional ordinary correctness defect was established in this pass.
+[Audit 01](01-correctness.md) and [audit 02](02-efficiency.md) retain their own
+findings. No implementation changes or additional consumer guarantees are
+implied by this refactoring catalog.

docs/roadmap/audit/05-resources.md

@@ -0,0 +1,215 @@
+# Resource Management Audit
+
+Reviewed 2026-11-04 against commit `fb87fab`.
+
+## Scope And Evidence
+
+Reviewed file and response-body ownership, bounded reads, HTTP client/transport
+lifetimes, timeout contexts and retry timers, backend admission and permits,
+prepared-handle cleanup, retained snapshots, JSON conversion, rendering buffers,
+schema resources, and repair feedback. The architecture uses immutable
+engine-lifetime configuration, invocation-local execution/validation resources,
+and caller-owned public results. Production code launches no PromptKit-owned
+goroutines or background workers requiring a shutdown protocol.
+
+Evidence consists of source/call-graph review, relevant existing tests, and a
+small temporary offline diagnostic outside the repository. The diagnostic used
+a controlled HTTP transport, counted body reads/closure, inspected the existing
+prepared-handle owner reference, and compared string backing addresses. No
+benchmark, heap-size measurement, deliberate exhaustion, live network call,
+production change, or new repository test was needed.
+
+Focused existing tests passed for content-file closure, HTTP error-body limits
+and closure, independent attempt contexts/body closure, read cancellation,
+prepared discard/ownership, diagnostic formatting, synchronous schema
+cancellation, and repeated-value JSON work bounds. These tests establish the
+existing protected behavior; they do not cover all findings below.
+
+## 1. Authentication Diagnostics Read An Unbounded Upstream Body
+
+**Priority:** P2. Unbounded remote-response consumption; a finite limit would be
+a new policy decision, not enforcement of an existing documented byte limit.
+
+**Code:** [runOpenRouterKey](../../../internal/usecase/smoke_authentication.go:66),
+[recognizedOpenRouterKeyResponse](../../../internal/usecase/smoke_authentication.go:139),
+and [diagnostic HTTP execution](../../../internal/llm/http_evidence.go:129).
+
+**Circumstances:** An authentication endpoint returns HTTP 200 with a large or
+continuously growing body. Recognition calls unrestricted `io.ReadAll` and then
+unmarshals the whole document into raw-message maps. Neither declared length nor
+streamed bytes are bounded on this path. Generation's 16 MiB success-body limit
+and 64 KiB error-body limit do not apply to diagnostic GETs.
+
+**Consequences / evidence:** The offline diagnostic supplied a valid small
+identifying `data` object plus 4 MiB of irrelevant padding. The probe read all
+4,194,375 bytes, closed the body, and reported credential acceptance. Increasing
+the padding has no corresponding byte-budget check. A fast misconfigured or
+unexpected endpoint can consume substantial memory before its timeout; concurrent
+authentication probes multiply consumption because HTTP-only diagnostics bypass
+generation admission. A timeout bounds elapsed time, not bytes allocated within
+that time. There is no demonstrated descriptor leak: the body is closed.
+
+**Potential remediation:** Select a finite body budget appropriate to this small
+diagnostic protocol, enforce it for both declared and streamed/underreported
+lengths, and close without unbounded draining on overflow. Preserve single-JSON
+framing and the distinction between unknown acceptance and observed rejection;
+an oversized response must not be reported as credential rejection. Decide the
+budget and overflow report classification explicitly, then update the owning
+contract and tests. This need not impose new limits on application prompts,
+artifacts, or generated output.
+
+## 2. Convenience Source Conversion Expands Shared Graphs Before Bounded Loading
+
+**Priority:** P2. Potentially exponential allocation before upstream resource
+checks can run.
+
+**Code:** [providerMapForMapping](../../../provider_values.go:19),
+[providerValueForMapping](../../../provider_values.go:31),
+[ProfileSource](../../../profiles.go:33), and
+[BackendSource](../../../backends.go:100).
+
+**Circumstances:** A caller builds a small acyclic Go value graph containing
+repeated map/slice references. Each level can point twice to the same child:
+
+```go
+value := map[string]any{"value": true}
+for range depth {
+    value = map[string]any{"left": value, "right": value}
+}
+```
+
+The source converter tracks only active ancestors to identify cycles. Once one
+branch finishes, its visit is removed, so the sibling independently converts
+the same entire child. It allocates a fresh expanded tree before calling
+`configkit.Mapping`. For d levels, O(d) distinct input maps become
+Theta(2^d) output maps. The depth guard does not constrain width or total work;
+this expansion can become impractical well below depth 100.
+
+**Consequences:** Programmatic source creation can exhaust memory before
+`NewEngine` or ConfigKit's deferred diagnostics are reached. This is not a claim
+that ConfigKit lacks bounds: the selected ConfigKit v1.1.0's mapping builder
+checks node, byte, and depth budgets during its own snapshot traversal
+(`internal/engine/value.go`, `valueBuilder.build`). It is called only after
+PromptKit has already expanded the graph. PromptKit's later `jsonvalue` node
+budget is also too late to protect this conversion.
+
+**Potential remediation:** Memoize completed structural conversions so repeated
+input containers reuse already-converted containers until ConfigKit snapshots
+them with its existing bounds. Keep active-ancestor tracking distinct from the
+completed memo so cycles still become deferred source errors. Account correctly
+for container identity, including distinct slice views. Alternatively, seek
+upstream support for the additional Go value forms so this conversion can be
+removed. Preserve immediate source ownership and deferred failure behavior;
+do not introduce a PromptKit raw-source validation pass or duplicate ConfigKit's
+configuration limits to fix this pre-handoff allocation problem.
+
+## 3. Inert Prepared Handles Retain Their Entire Execution Owner
+
+**Priority:** P3. Avoidable retained-memory dependency after use/discard.
+
+**Code:** [PreparedExecution.owner](../../../internal/usecase/prepared_execution.go:24),
+[Discard](../../../internal/usecase/prepared_execution.go:96),
+[claim](../../../internal/usecase/prepared_execution.go:172), and
+[payload clearing](../../../internal/usecase/prepared_execution.go:188).
+
+**Circumstances:** A consumer retains a completed or discarded handle for its
+supported `Details` access, while replacing or dropping the originating engine.
+Payload clearing releases the per-execution plan, credential, observer, and
+execution snapshot, but the handle's `owner *Runner` remains unchanged.
+
+**Consequences / evidence:** The reference path is
+`handle -> Runner -> captured repositories/clients/validator/capacity manager`.
+Thus a small inert handle keeps the originating engine's complete runtime
+snapshot and collaborators reachable, not just its deliberately retained details.
+The offline diagnostic confirmed the same nonnil runner pointer after both
+discard and successful execution. Retaining handles from successive engine
+reconstructions can retain successive large snapshots. Keeping details itself
+is required behavior, and this has no additional owner-retention cost while the
+application still retains the same engine anyway.
+
+**Potential remediation:** Clear the owner reference when the handle becomes
+inert, checking ownership under the same mutex to avoid racing the current
+unlocked owner read. Another option is a small standalone identity token shared
+by a runner and its handles, which establishes ownership without retaining the
+runner graph. Such a token must have reliable distinct identity; do not rely on
+distinct zero-sized allocations comparing unequal. Preserve related namespace
+views, foreign-owner rejection, single claim, discard/run races, and details
+availability. No engine `Close` method or automatic closing of caller-owned
+collaborators is implied.
+
+## 4. Repair Feedback's Output Bound Does Not Bound Its Temporary Serialization
+
+**Priority:** P3. Avoidable memory amplification on validation failures.
+
+**Code:** [formatRepairDiagnostics](../../../internal/usecase/repairer.go:119).
+
+**Circumstances:** A failed schema validation produces a large diagnostic or many
+diagnostics and repair is enabled. The formatter normalizes every string, marshals
+the complete array, then discovers that it exceeds 64 KiB. It subsequently
+marshals individual full entries again before truncating one of them.
+
+**Consequences:** Although the returned feedback is bounded correctly, temporary
+storage scales with the complete diagnostic text, including text omitted from
+the request. JSON escaping can amplify its size, and normalization adds another
+copy for invalid UTF-8. Large validation messages and concurrent repairs can
+therefore create large additional buffers despite producing tiny feedback.
+The complete diagnostics already exist for the validation result; these extra
+full encodings do not help preserve them. Existing formatting tests check output
+bounds/data preservation, not scratch-space use.
+
+**Potential remediation:** Encode feedback incrementally with bounded buffering
+and lookahead sufficient to distinguish a complete array from an omitted suffix.
+Stop encoding omitted text rather than allocating its full JSON representation.
+Keep the existing 64 KiB result bound, ordered entries, rune-safe truncation,
+valid UTF-8, JSON escaping, and in-bound omission notice. Leave the original
+validation diagnostics and full previous candidate unchanged. This is an
+implementation opportunity, not a claim that the existing contract promises a
+64 KiB peak heap footprint or permission to add a diagnostic-input limit.
+
+## 5. Short Normalized Session IDs Can Pin Large Input Strings
+
+**Priority:** P3. Avoidable backing-buffer retention in retained metadata.
+
+**Code:** [NormalizeSessionID](../../../internal/domain/session.go:10),
+[preparation](../../../internal/usecase/runner.go:312), and
+[prepared metadata construction](../../../internal/usecase/runner.go:491).
+
+**Circumstances:** A direct or rendered session ID contains substantial leading
+or trailing whitespace around a short valid identifier. `strings.TrimSpace`
+returns a substring; normalization validates that substring and returns it
+without detaching its backing storage. Public conversion preserves the string.
+
+**Consequences / evidence:** An offline public-API probe prepared a request with
+1 MiB of spaces followed by `s`. The returned session ID had length one but its
+backing address still pointed inside the large original allocation. Retaining a
+prepared value, handle details, or run result can therefore retain that buffer
+after the original request/rendering buffer would otherwise be collectible.
+The 256-code-point semantic limit does not prevent this retention. This does not
+require changing which padded identifiers are accepted.
+
+**Potential remediation:** Detach accepted normalized session metadata with a
+small owned string, for example `strings.Clone` after validation. Preserve empty
+omission, Unicode trimming/counting, hashes, and wire content. Review other
+short canonical metadata returned by trimming for the same pattern when making
+the change; do not indiscriminately clone large prompt/message payloads or add a
+new raw-session-size restriction.
+
+## Cleanup Checks And Non-Findings
+
+No additional descriptor, timer, connection-ownership, or goroutine leak was
+established. Artifact/schema handles use deferred closure; prompt content handles
+close after reading; generation response bodies close on success and decoding,
+status, and size failures. Diagnostic bodies close, and their timeout cancellation
+is attached to body closure. Retry timers stop on exit. Admission/active permits
+release on errors, cancellation, and panic unwinding. Prepared payloads otherwise
+clear as intended.
+
+Caller-selected artifact/schema/prompt sizes, intentionally unlimited backends,
+and caller-retained ready handles are not classified as bugs merely because
+their total memory can grow. Synchronous dependency calls may delay cancellation
+by design, rather than being detached into leaking goroutines. Borrowed HTTP
+transports/filesystems/clients remain subject to their ownership contracts; this
+audit proposes no blanket shutdown or disposal API.
+
+Previously recorded repeated-copying issues remain in
+[the efficiency audit](02-efficiency.md); they are not duplicated here.

docs/roadmap/audit/06-error-handling.md

@@ -0,0 +1,197 @@
+# Error-Handling Audit
+
+Reviewed 2026-11-04 against commit `6b1323d`.
+
+## Scope And Evidence
+
+Reviewed error detection, wrapping, public translation, diagnostic reporting,
+cancellation and timeout handling, retries, validation/repair failures, and
+cleanup across construction, configuration inspection, prompt/profile selection,
+artifact reading, rendering, ordinary and prepared execution, generation, and
+smoke diagnostics, plus the small release-entrypoint wrappers. Contracts were
+checked against GoDoc, integration references, the architecture policy, ADR 0001,
+and existing tests.
+
+Configuration composition belongs to ConfigKit; PromptKit validates and
+interprets its final handoff and preserves upstream load errors. The runner adds
+operation context, generation coordinates attempts and status retries, the HTTP
+adapter classifies/redacts provider failures, and the public facade translates
+internal categories. Content-validation failures and concluded diagnostic
+failures are generally data rather than Go errors under their respective
+contracts. Prepared handles permit one execution claim; failure after a
+successful claim does not make the handle reusable.
+
+Evidence includes source/call-graph review and a temporary public-API diagnostic
+outside the repository. It used injected collaborators and a controlled HTTP
+transport; it made no live network calls. No production code or repository
+tests were changed.
+
+Existing error, failure, cancellation, timeout, panic-cleanup, and retry tests
+passed using:
+
+```sh
+go test -count=1 -run 'Error|Failure|Fail|Cancel|Timeout|Panic|Release|Retry|Retries' \
+  . ./internal/llm ./internal/generation ./internal/usecase \
+  ./internal/capacity ./internal/validate ./internal/artifact \
+  ./internal/configuration
+```
+
+These tests establish protected behavior but do not cover the three reproductions
+below. Findings from the earlier correctness and resource-management audits are
+not repeated here.
+
+## 1. Nested Collaborator Errors Can Hide The Owning Operation's Public Category
+
+**Priority:** P2. Confirmed public error-classification defect.
+
+**Code:** [publicErrorFor](../../../errors.go:63),
+[mapPublicError](../../../errors.go:16), and
+[wrapGenerationError](../../../internal/usecase/runner.go:271).
+
+**Failure condition:** An injected `LLMClient` or `ArtifactReader` returns an
+error obtained from another PromptKit operation. Ordinary mapped errors retain
+their internal chain. `publicErrorFor` searches that complete chain with
+`errors.Is` and chooses the first matching category in a fixed priority order,
+regardless of which operation the outer runner classified.
+
+**Expected:** A model-client failure matches `ErrLLMGenerate`; an artifact-reader
+failure matches `ErrArtifactLoad`. The injected cause remains available through
+`errors.Is`, as their [public GoDoc](../../../engine.go:49) specifies. Preserving
+the nested cause may also preserve its public categories; it must not replace
+the category for the operation that just failed.
+
+**Actual / reproduction:** A child engine's `Run` returns a missing-prompt error.
+A parent engine selects a valid prompt, then its injected client returns that
+child error. The outer error contains the text `failed to generate output`, but:
+
+```text
+errors.Is(err, ErrLLMGenerate)    = false
+errors.Is(err, ErrPromptNotFound) = true
+errors.Is(err, childError)        = true
+```
+
+Returning the same child error from an injected artifact reader similarly
+produces `errors.Is(err, ErrArtifactLoad) == false`. This is reproducible using
+only public APIs, without importing internal sentinels. Consumers can therefore
+miss the documented failure category when composing PromptKit-backed adapters.
+
+**Potential remediation:** Translate the owning operation's classification
+independently of an arbitrary collaborator's nested chain, while retaining the
+original injected cause. Explicit boundary/stage information or mapping at the
+operation boundary can do this. Merely reordering the global switch risks
+breaking intentional direct-selection behavior: a genuinely missing prompt or
+profile must retain its documented distinction from other load failures.
+Protect nested-client and nested-reader cases through public-boundary regression
+coverage; preserve existing capacity, retry-exhaustion, provider-error redaction,
+and context identities.
+
+## 2. Successful-Status Body Read Failures Are Misreported As Malformed JSON
+
+**Priority:** P2. Confirmed loss of failure provenance and cause.
+
+**Code:** [decodeOpenAIChatResponse](../../../internal/llm/openai_compatible_client.go:373).
+
+**Failure condition:** A 2xx response body returns a non-EOF I/O error other than
+`context.Canceled` or `context.DeadlineExceeded`, either during the first decode
+or while checking the required final EOF. For example, a connection can fail
+while reading the response stream after headers were received.
+
+**Expected:** Distinguish an underlying read/transport failure from evidence of
+invalid JSON or trailing data. Retain the read cause through a redacting error
+so callers can inspect it with `errors.Is`/`errors.As`, consistent with the
+[transport category](../../../engine.go:65). Do not report trailing data when the
+reader failed without supplying any. This does not change the established
+classification of invalid/truncated JSON or introduce transport retries.
+
+**Actual / reproduction:** The controlled transport returned HTTP 200 and a body
+that supplied either a partial JSON prefix or a complete valid chat envelope,
+then returned a distinct read-error sentinel. Both executions returned
+`ErrLLMGenerate` plus `ErrMalformedResponse`, did not match `ErrTransport`, and
+discarded the sentinel. For the complete envelope, the error said `response
+contains trailing data`, although no trailing bytes were supplied. Inference
+smoke diagnostics likewise reported `malformed_response` with nil Go error
+rather than a transport failure. Thus the issue affects both public execution
+errors and factual diagnostic classification.
+
+The integration contract explicitly guarantees exact causes for
+[`http.Client.Do` failures](../../integrations/openai-compatible-chat.md#response-handling),
+not every decoder error. The finding concerns the separate loss and
+misclassification of an actual body-read failure, rather than asserting a new
+requirement to expose JSON parser errors or provider content.
+
+**Potential remediation:** Track underlying reader failures separately from
+decoder/framing failures, or reliably distinguish those classes before replacing
+the decode error. Reuse the redacting transport-error path for genuine read
+failures. Keep syntax/type errors, EOF-truncated JSON, extra JSON values, and
+size overflow in the existing malformed-response category; preserve cancellation
+precedence and bounded reads/closure. Add focused coverage for interrupted
+initial decoding and interrupted EOF verification, including diagnostic failure
+classification. Default formatting must remain free of raw response content and
+private transport details.
+
+## 3. Inference Diagnostics Swallow Some Invalid Request-Construction Errors
+
+**Priority:** P2. Confirmed mismatch with the diagnostic error contract.
+
+**Code:** [runInference](../../../internal/usecase/smoke_tester.go:207),
+[generation preparation](../../../internal/generation/coordinator.go:96), and
+[request payload validation](../../../internal/llm/openai_compatible_client.go:511).
+
+**Failure condition:** An inference diagnostic's effective settings are invalid
+in a way detected by built-in request construction rather than target selection.
+A runtime `Execution.ExtraParams` override containing the reserved key `model`
+is one public example. The coordinator returns `llm.ErrInvalidRequest` before
+dispatch with no attempt evidence. `runInference` treats every generation error
+as a concluded diagnostic failure.
+
+**Expected:** The [SmokeTest contract](../../../smoke_diagnostics.go:238) says invalid
+kinds, settings, and selections return an ordinary Go error before diagnostic
+work. A locally invalid request should follow that path and expose the public
+invalid-request category. This is distinct from a provider rejecting a validly
+constructed request or a diagnostic concluding with an observed failure.
+
+**Actual / reproduction:** Calling text-inference `SmokeTest` with
+`Execution.ExtraParams: map[string]any{"model": "override"}` returned nil Go
+error and a report with `Completed == true`, `Failure == invalid_request`, no
+attempts, and zero transport calls. The request-construction error and its
+explanation were discarded. The caller cannot use the documented ordinary-error
+path to identify this configuration mistake.
+
+**Potential remediation:** Preserve the distinction between local built-in
+request-preparation failures and concluded execution failures. Return preparation
+errors through the use-case/public invalid-request mapping rather than marking
+the diagnostic complete; preparation can remain in the generation adapter, with
+enough provenance returned to the caller to make that distinction. Do not infer
+that every zero-attempt diagnostic is an error: missing required credentials,
+capacity rejection, unsupported operations, and inapplicability have intentional
+report semantics. Preserve concluded provider failures and the separate
+caller-cancellation behavior. Add one public-boundary regression for the
+reserved-field override, retaining existing tests for those intentional reports.
+
+## Reviewed Behavior Not Raised As Defects
+
+- ConfigKit owns raw-source composition and errors before the accepted handoff;
+  no additional PromptKit raw-source validation is proposed.
+- Non-2xx provider detail parsing is deliberately best-effort. Unreadable,
+  oversized, or unrecognized error bodies retain the received status without
+  exposing the raw body. That differs intentionally from decoding a 2xx body.
+- Retries are limited to selected HTTP statuses. Transport failures, malformed
+  responses, and arbitrary injected-client failures are not automatically
+  retried; no expansion is needed for these findings.
+- Completed output-validation failures are returned as validation results;
+  operational schema failures are errors. Repair failure propagation and
+  exhaustion retain that distinction.
+- Caller cancellation, configured request timeouts, cancellation during capacity
+  acquisition/backoff, synchronous schema operations, and partial diagnostic
+  evidence have explicit, tested handling. Callback panics propagate while
+  deferred permit/admission cleanup still runs.
+- Ignored SHA-256 writer errors cannot occur with the selected implementation.
+  Best-effort closure of read-only files and HTTP bodies does not imply a
+  persistence guarantee or require converting a completed read into a failure.
+- Release wrappers use fail-fast shell execution and `exec` to propagate the
+  selected tool's status. Publication and recovery implementation belongs to
+  ReleaseKit; no release or publication operation was run for this audit.
+
+No additional state-recovery or cancellation defect was substantiated in this
+pass. The findings do not call for a new recovery subsystem, automatic replay of
+possibly completed provider work, or changes to source ownership.

docs/roadmap/audit/07-security.md

@@ -0,0 +1,205 @@
+# Security Audit
+
+Reviewed 2026-11-04 against commit `cf0bfd4` and the current worktree. Pending
+relocations of earlier audit documents do not change the production baseline
+and were left untouched.
+
+## Scope, Architecture, And Threat Boundary
+
+PromptKit is an in-process Go library with a public facade and internal
+configuration, execution, transport, validation, and observation components.
+ConfigKit owns source loading/composition; PromptKit interprets its final
+handoff and captures immutable definitions. Execution reads caller-selected
+artifacts, renders captured templates, resolves credentials, and contacts the
+selected backend. Prepared handles privately retain one execution's settings.
+Inspection and observation expose deliberately selected metadata.
+
+Reviewed configuration and request validation, filesystem/resource selection,
+template and JSON handling, credential flow, public formatting and reports,
+HTTP requests/redirect handling, provider responses, schema loading, repair,
+backend capacity, prepared ownership, and release-entrypoint wiring. The review
+included interactions with the selected ConfigKit and JSON Schema dependencies
+and an advisory scan of runtime packages.
+
+The [architecture policy](../../policy/architecture.md#repository-and-consumer-boundary)
+and [consumer security guidance](../../consumers/pkg-promptkit.md#protect-files-and-generated-data)
+place application authorization, untrusted path/endpoint admission, payload-size
+policy, and output disclosure with the consumer. Configuration sources,
+templates, and injected Go collaborators are application-selected capabilities.
+Upstream HTTP response bytes remain an external trust boundary. This review
+does not treat ordinary file selection, runtime endpoint overrides, namespace
+selection, or supported message roles as authorization mechanisms.
+
+No production code or repository tests were changed, no secrets were collected,
+and no external provider or forge was probed. Network activity was limited to
+obtaining the public Go vulnerability checker/database and reading official
+advisories. This is a source review with focused diagnostics, not a claim that
+every possible consumer deployment is secure.
+
+## Evidence
+
+- Source/call-graph review and the applicable public GoDoc, framework formats,
+  integration contract, policies, and ADR 0001.
+- Existing focused tests passed for credential/report formatting, sensitive
+  configuration inspection/diffs, schema reference containment, provider-body
+  bounds, JSON value cycles/depth, prepared ownership, and capacity boundaries.
+- A temporary offline public-API diagnostic used a controlled HTTP transport
+  and a synthetic credential. It verified the authentication-body behavior in
+  finding 2 and confirmed that credential was absent from the serialized report.
+- `govulncheck` v1.9.0 scanned `./...` using Go 1.27.0 and the Go vulnerability
+  database last modified `2026-11-02T22:01:43Z`. The runtime scan covered the
+  module, its examples, ConfigKit v1.1.0, jsonschema/v6 v6.0.2, YAML v4.0.0-rc.6,
+  and x/text v0.14.0. It is not a complete audit of ReleaseKit's implementation
+  or every dependency used only by the release tool.
+
+The text-mode scan reported ten symbol-level standard-library advisories, plus
+package/module-only matches. Those are candidate matches, not ten established
+PromptKit vulnerabilities. Applicability was checked against the actual
+operations described below rather than inferred from interface-call traces.
+
+## 1. The Audited Go Toolchain Predates Applicable Networking Security Fixes
+
+**Priority:** P2. Build/deployment security finding, not a newly discovered
+PromptKit algorithm defect.
+
+**Relevant code:** [declared Go minimum](../../../go.mod:3),
+[default HTTP-client construction](../../../internal/llm/openai_compatible_client.go:58),
+[generation dispatch](../../../internal/llm/openai_compatible_client.go:211), and
+[diagnostic dispatch](../../../internal/llm/http_evidence.go:129).
+
+**Condition and impact:** The local toolchain used for this audit is Go 1.27.0.
+Applications built with that patch can exercise affected standard-library
+networking through PromptKit's normal outbound client:
+
+- [GO-2026-6611](https://pkg.go.dev/vuln/GO-2026-6611) describes excessive CPU
+  consumption caused by a malicious HTTP/2 peer repeatedly changing initial
+  stream windows. An application using HTTP/2 against a hostile or compromised
+  backend, with enough active streams for amplification, can suffer availability
+  loss. The Go 1.27 fix is in 1.27.2. Body-size limits do not address protocol
+  processing below response decoding.
+- [GO-2026-6809](https://pkg.go.dev/vuln/GO-2026-6809) adds explicit rejection of
+  compromised intermediate certificates. The affected conditions depend on the
+  platform and trust-store state. When the old trust path remains accepted, an
+  attacker able to present a matching certificate chain and intercept the
+  connection could compromise confidentiality/integrity of prompts, responses,
+  or bearer credentials. The Go 1.27 fix is in 1.27.3. This is not a claim that
+  every default system trust store remains affected.
+
+**Evidence and limits:** The advisories establish affected versions and attack
+conditions; source review establishes use of the standard outbound HTTP/TLS
+path. No certificate forgery, HTTP/2 attack, or running consumer binary was
+tested. The scanner's surprising interface-based paths from unrelated methods
+are not relied upon as proof of exploitation. PromptKit is source-only: a
+consumer's actual compiled toolchain and transport configuration determine its
+exposure. The `go 1.27.0` directive is a compatibility minimum, not a record of
+which patch built an application.
+
+**Potential remediation:** Use an appropriate current patched Go toolchain,
+at least 1.27.3 within this series for these two fixes, for release validation
+and ReleaseKit execution; advise consumers to rebuild affected applications
+with patched Go. Recheck advisories against that selected toolchain. This fits
+the existing distinction between consumer minimum and maintainer toolchain in
+the [release procedure](../../release.md#prepare-the-candidate).
+Whether to raise the consumer minimum for the identified fixes is a separate
+compatibility decision under the existing policy, not an automatic consequence
+of this audit. These fixes do not require custom TLS verification, a blanket
+HTTP/2 ban, or a new recurring release gate.
+
+## 2. Authentication Probes Permit Unbounded Remote-Body Memory Consumption
+
+**Priority:** P2. Existing availability finding, confirmed at the external
+response boundary. This is the same mechanism as resource-management audit
+finding 1, not a second independent remediation item.
+
+**Relevant code:** [authentication recognition](../../../internal/usecase/smoke_authentication.go:139),
+[authentication probe execution](../../../internal/usecase/smoke_authentication.go:66),
+and [diagnostic HTTP execution](../../../internal/llm/http_evidence.go:129).
+
+**Condition:** A consumer explicitly invokes an `openrouter_key` authentication
+diagnostic. The selected endpoint, a compromised provider, or an intermediary
+on an operator-selected plaintext connection returns HTTP 200 with a large or
+continuously growing body. Recognition uses unrestricted `io.ReadAll` followed
+by whole-document JSON unmarshalling. Unlike generation response decoding,
+this path has no byte budget. HTTP-only probes also do not use generation
+capacity admission.
+
+**Impact:** The remote peer controls memory consumption before the diagnostic
+returns. A fast response can cause large allocations or process termination
+before a deadline, and concurrent probes multiply the exposure. A configured
+timeout bounds time, not bytes allocated during that interval. This requires a
+diagnostic call; it is not an unauthenticated inbound attack on a PromptKit
+server, and ordinary `Run` does not execute the probe.
+
+**Evidence:** The offline diagnostic returned a minimal recognized account
+object and 4 MiB of irrelevant padding. PromptKit read all 4,194,374 bytes,
+closed the body, and concluded with credential acceptance. The body used an
+unknown content length. No deliberate memory exhaustion was attempted. The
+synthetic bearer credential was supplied to the controlled transport and was
+absent from the serialized diagnostic report.
+
+**Potential remediation:** Select a finite body budget for this small diagnostic
+protocol and enforce it for declared and streamed lengths, closing on overflow
+without unbounded draining. Preserve factual reporting: overflow establishes
+neither credential rejection nor acceptance. Agree the limit and overflow
+classification explicitly; no existing authentication-body byte guarantee is
+being enforced. The existing resource-audit remediation can own this fix;
+there is no need for separate global limits on prompts or artifacts.
+
+## Other Candidates And Boundary Checks
+
+- **Scanner applicability:** Server-only advisories GO-2026-6603, 6612, 6613,
+  and 6617 do not establish an attack on PromptKit's outbound client. ECH
+  server parsing (6607), multipart-form parsing (6608), and Range-serving
+  behavior (6609) likewise lack a matching operation here. The CONNECT-with-body
+  condition in 6605 and raw HTTP/2-to-HTTP/1 header forwarding in 6610 are absent
+  from built-in POST/GET generation and factual diagnostic reporting. Windows
+  junction-directory creation (6604) is not established by read-only resource
+  opens. These exclusions concern PromptKit, not all possible code in a host
+  application.
+- **Module-only matches:** x/text advisories GO-2026-5970 and 6629 were not
+  reported as called vulnerable symbols; no matching use of `norm.Iter` or
+  `precis.Nickname` was established. html/template advisories 6599 and 6600 do
+  not apply to the text/template renderer. No separate runtime third-party
+  vulnerability was substantiated by this scan; that is not a general safety
+  certification of those modules.
+- **Files and schemas:** Ordinary artifacts intentionally allow unrestricted
+  caller-selected regular files. Host-backed `content_file` contexts use
+  ConfigKit's rooted filesystem, while the standard schema loader checks
+  canonical containment and rejects remote reference schemes. Generic `fs.FS`
+  inputs remain caller-supplied capabilities; logical path checks do not turn an
+  arbitrary filesystem implementation into a host sandbox. No additional
+  confinement guarantee is proposed.
+- **Injection:** Variables and artifact text are rendered as data rather than
+  reparsed as Go templates. Captured templates have the input helper and normal
+  text/template functionality, not a shell or process-execution helper.
+  Provider payloads use JSON encoding and reserved-field checks. Model output
+  is data for validation/repair, not executable code or an authorization
+  decision. Prompt injection and domain-aware interpretation remain consumer
+  concerns.
+- **Credentials and disclosure:** Direct keys are request-scoped, excluded from
+  supported JSON and public request summaries, and removed from diagnostic
+  settings and prepared details. Provider error defaults redact details;
+  explicit provider accessors and arbitrary application content retain the
+  documented consumer disclosure responsibility. Inspection/diff tests protect
+  configured sensitive provider paths. No blanket redaction of arbitrary
+  prompt, schema, variable, error, or provider content is promised.
+- **Ownership and availability:** Engine snapshots and copied public values
+  prevent ordinary cross-call mutation. Prepared handles enforce owner binding
+  and single claim. Capacity and retry cleanup are tested. Intentionally
+  unlimited backends, synchronous template/schema dependency calls, and
+  application-selected large files are not newly classified as vulnerabilities.
+  The other resource-audit findings remain maintenance items rather than
+  independently demonstrated remote attacks.
+- **Release tooling:** Local launchers pass quoted arguments through `exec`;
+  there is no PromptKit runtime command-execution surface. The delivery token
+  path is ignored and selected for committed-file protection. Publication,
+  token-file permission enforcement, and recovery implementation belong to
+  ReleaseKit; no publication command was run during this audit.
+
+## Recommended Follow-Up
+
+Use a patched Go toolchain before release preparation and address the existing
+authentication-body availability finding. Neither requires a new application
+authorization subsystem or stronger filesystem, endpoint, or semantic-output
+contracts. No additional PromptKit-specific confidentiality, integrity, or
+authorization defect was substantiated within the documented boundary.

docs/roadmap/audit/08-testing.md

@@ -0,0 +1,284 @@
+# Test Coverage And Effectiveness Audit
+
+Reviewed 2026-11-04 against commit `f7085d7` and the current worktree. Existing
+audit-document moves were preserved. No production code or tests were changed.
+
+## Assessment And Method
+
+The suite broadly follows the [testing policy](../../policy/testing.md): real
+ConfigKit composition, filesystems, rendering, validation, and engine assembly
+are exercised alongside controlled model clients and HTTP transports. Internal
+tests own focused rules and coordination, external-package root tests protect
+the public facade, and the three maintained examples have Go workflow tests.
+Capacity, prepared-handle state transitions, ownership, cancellation, retries,
+repair, observations, and configuration inspection have substantial behavioral
+coverage. This is a targeted improvement list, not a recommendation to replace
+the test architecture or pursue a percentage threshold.
+
+Both required analyses were performed: review of missing behavioral protection,
+and review of assertion/fixture cost and coupling. Evidence includes source and
+call-graph review, existing GoDoc and integration contracts, full-suite coverage,
+repeated race-enabled execution, and reuse of the earlier error audit's temporary
+offline diagnostic. No live provider, paid API, new repository test, production
+mutation, or external mutation-testing service was needed.
+
+Validation used the local Go 1.27.0 toolchain:
+
+```sh
+go test -count=1 -covermode=atomic -coverpkg=./... -coverprofile=/tmp/promptkit-testing-audit-LHAyNm/coverage.out ./...
+go tool cover -func=/tmp/promptkit-testing-audit-LHAyNm/coverage.out
+go test -race -vet=off -shuffle=42 -count=2 ./...
+```
+
+Both suite runs passed, including the examples. The race run exercised two
+shuffled passes; this is evidence against readily reproducible order/race
+problems, not proof that no flakes exist. Coverage output remains outside the
+repository. The commands are audit evidence, not additional release gates.
+
+Combined coverage is **86.4%**, or 4,782 of 5,536 instrumented statements.
+Selected package totals below combine hits across all test binaries. The
+per-binary percentages printed by `-coverpkg=./...` are not individual-package
+coverage percentages; repeated profile spans were combined before calculating
+these totals.
+
+| Component | Combined statement coverage |
+| --- | ---: |
+| Public facade | 81.0% |
+| Configuration | 91.0% |
+| Domain | 98.0% |
+| Capacity | 98.0% |
+| Generation coordination | 91.2% |
+| HTTP model client | 94.2% |
+| Execution use cases | 90.5% |
+| Rendering and requirement inspection | 78.4% |
+| Output/schema validation | 75.8% |
+
+Coverage located candidates; the findings below concern specific risks and
+assertions, not the percentages themselves. Testing recommendations associated
+with already recorded implementation defects should accompany those fixes,
+rather than becoming duplicate remediation projects.
+
+## 1. RunRequest's Privacy Formatting Contract Is Not Exercised
+
+**Priority:** P2. Missing protection for a documented confidentiality boundary.
+
+**Relevant code:** [RunRequest formatting](../../../formatting.go:5) and the
+analogous existing [SmokeTestRequest privacy test](../../../smoke_test_contract_test.go:135).
+
+**Deficiency:** `RunRequest.String`, `GoString`, and their shared formatter all
+have 0% combined statement coverage. The suite protects formatting of generated
+client requests, smoke-test requests, and opaque prepared handles, but does not
+exercise the ordinary run request's own formatting contract.
+
+**Unverified behavior:** A change that prints the direct key, input bodies, or
+variable values from `RunRequest` can pass the complete suite. This is an
+assertion gap, not evidence that current formatting actually leaks those values.
+
+**Potential improvement:** Add one small public-boundary test with distinct
+synthetic private markers and exercise the supported String/GoString formatting
+paths on values and pointers. Assert absence of the data covered by the GoDoc;
+do not snapshot the complete summary or freeze incidental field order/count
+formatting. If adding a JSON check, assert only the documented direct-key
+exclusion, not a stable JSON shape or omission of ordinary request data.
+Reflection-based disclosure is explicitly outside this formatter guarantee.
+
+## 2. The Public Single-Schema-File Option Has No Execution Coverage
+
+**Priority:** P2. An exported source-selection path is entirely unprotected.
+
+**Relevant code:** [WithSchemaFile](../../../engine.go:212),
+[SetSchemaFile](../../../internal/engine/engine.go:75), and
+[fileSource](../../../internal/engine/engine.go:211).
+
+**Deficiency:** All three functions have 0% combined coverage. The existing
+[single-file validator test](../../../internal/validate/standard_validator_test.go:596)
+constructs `FSValidator` directly with `fstest.MapFS`; it does not test the public
+option, host-file selection, or its engine wiring.
+
+**Unverified behavior:** The public option could ignore its argument, fail to
+replace `SchemaDir`, select the wrong root/base name, or mishandle invalid files
+without this validator test noticing. Direct validator success does not establish
+that a consumer can reach that behavior through the supported API.
+
+**Potential improvement:** Use `t.TempDir` and a real schema file in a small
+public integration test. Exercise construction plus preparation/validation
+through that file's base name, and representative invalid selections such as a
+missing file or directory. Include source-option replacement where it protects
+the documented last-valid-option behavior. Assert public error identity and
+observable schema selection, not private helper calls. This does not require a
+helper-by-helper test matrix or model-provider access.
+
+## 3. Programmatic Provider-Value Conversion Lacks Representative Value Cases
+
+**Priority:** P2. Conversion and ownership coverage misses known data-integrity
+and deferred-error risks.
+
+**Relevant tests:** [programmatic source ownership/order](../../../sources_contract_test.go:14)
+and [programmatic construction](../../../construction_contract_test.go:226).
+**Implementation:** [providerValueForMapping](../../../provider_values.go:31).
+
+**Deficiency:** The converter has only 11.9% combined coverage. The principal
+source-ownership test mutates a flat string-valued map. It does not establish
+preservation and ownership for representative nested/typed containers or the
+`json.Number` branch, nor deferred failure for unsupported provider values.
+The more extensive `internal/jsonvalue` tests protect a different traversal and
+do not execute this pre-ConfigKit convenience conversion.
+
+**Unverified behavior:** The suite passes with the exact-number rounding defect
+already recorded in [the correctness audit](01-correctness.md#1-provider-source-helpers-silently-change-exact-json-numbers).
+It also would not detect many regressions in nested source capture or the
+documented source-creation-versus-loading failure boundary.
+
+**Potential improvement:** Exercise both `ProfileSource` and `BackendSource`
+through `NewEngine` with a compact set of representative supported nested
+values, mutations after source creation, and unsupported/cyclic values that
+should fail during loading. Add the numeric regression when addressing the
+existing precision finding. Compare meaningful final values or encoding, not Go
+container identity, numeric spelling, or a required float representation.
+Keep tests at the final handoff/public source-helper boundary; do not duplicate
+ConfigKit's raw-source validation or its entire value-limit suite. Work-bound
+coverage for the separately identified shared-graph issue belongs with that fix.
+
+## 4. Error Tests Miss Composed Causes And Pre-Dispatch Diagnostic Failures
+
+**Priority:** P2. Covered code paths lack important boundary-case assertions.
+
+**Relevant tests:** [public generation failures](../../../generation_error_contract_test.go:111),
+[direct mapper cases](../../../errors_internal_test.go:13),
+[body cancellation](../../../internal/llm/prepared_generation_test.go:161), and
+[inference diagnostic cases](../../../internal/usecase/smoke_inference_test.go:14).
+
+**Deficiency and demonstrated blind spots:**
+
+- Injected failures are mostly plain marker errors or a single explicit
+  category. They do not check a real nested PromptKit error returned through a
+  different operation. The suite passes although an injected client/reader can
+  lose the outer public generation/artifact category.
+- Tests cover `http.Client.Do` failures and cancellation during successful-body
+  reading, but not an ordinary non-EOF read error during initial decoding or
+  final EOF verification. Such errors currently lose their cause and become
+  malformed-response failures.
+- Diagnostic tests cover invalid selection/settings and concluded generation
+  failures separately, but miss a runtime setting rejected by built-in request
+  construction before dispatch. A reserved `ExtraParams["model"]` override
+  currently returns a completed report with nil Go error.
+
+These are the three implementation findings in the error-handling audit, not
+new duplicate defects. Re-running its temporary offline public-API diagnostic
+against the current tree confirmed all three while the complete suite passed.
+
+**Potential improvement:** Add narrowly owned regressions with those fixes:
+real nested public causes through the injection boundary; a controlled body
+reader failing at each decode/framing phase; and a public `SmokeTest` call with
+a reserved runtime field and counted transport. Assert relevant error identities,
+cause preservation, factual failure classification, and whether dispatch occurred.
+Avoid reproducing the internal error topology by hand or matching complete
+messages. Preserve intentional nil-error diagnostic findings, cancellation
+precedence, best-effort non-2xx detail parsing, and status-only retry policy.
+
+## 5. Multi-Resource Schema Tests Do Not Establish A Usable Provider Constraint
+
+**Priority:** P2. Missing assembled validation-to-wire assertion.
+
+**Relevant tests:** [prepared execution source freezing](../../../prepared_execution_contract_test.go:19),
+its [referenced schema fixture](../../../prepared_execution_contract_test.go:809),
+and [local reference validation](../../../internal/validate/standard_validator_test.go:627).
+
+**Deficiency:** Local tests establish that referenced schemas compile and
+validate. The prepared-execution test checks that the injected client's schema
+matches frozen details. Its opaque model client accepts that schema without
+interpreting references. Equality between those two values does not establish
+that the provider-facing constraint contains the resources it needs.
+
+**Unverified behavior:** Both sides can contain the same unresolved `value.json`
+reference while the test passes. This is the packaging gap already identified
+in [the correctness audit](01-correctness.md#2-external-schema-resources-do-not-reach-the-provider-constraint).
+The freezing test's fake is appropriate for its stated ownership purpose; the
+missing protection is a separate built-in-path integration case.
+
+**Potential improvement:** Once the schema-packaging policy is selected, add one
+offline assembled test with a root and child schema, capture the actual built-in
+HTTP payload, and assess the chosen provider-facing representation independently
+of `Details`. For a bundled native schema, compile/evaluate the captured
+constraint without granting access to PromptKit's original local files; if the
+chosen policy supports a local-only fallback, assert that behavior instead.
+Keep the existing freezing test. Do not require live provider acceptance or
+promise support for arbitrary provider schema dialect features.
+
+## 6. Artifact Tests And Benchmarks Pin An Explicitly Opaque Hash Algorithm
+
+**Priority:** P3. Remove or rewrite change-detector assertions, retaining the
+substantive file-reading tests.
+
+**Relevant assertions:** [size-hint test](../../../internal/artifact/reader_test.go:287)
+and [benchmark correctness check](../../../internal/artifact/reader_benchmark_test.go:156).
+
+**Deficiency:** Both independently calculate a hex SHA-256 digest and require
+that exact value. The [public Artifact.Hash contract](../../../types.go:290)
+expressly leaves its format and algorithm unspecified. A legitimate algorithm
+or encoding change would fail tests unrelated to the requested behavior.
+
+**Potential improvement:** Replace those algorithm-specific checks with the
+opaque equality relationships needed for each test. For the under/overreported
+size cases, comparing with the inline reader for the same complete body still
+detects hashing only a reported-size prefix. The existing
+[source-parity test](../../../internal/artifact/reader_test.go:27) already owns
+repeatability, same-content equality, and changed-content inequality. Retain
+body, actual-size, metadata, and read-boundary assertions. The benchmark may
+reuse an opaque expected hash prepared outside measurement or rely on the
+owning behavioral test; it need not maintain its own SHA-256 contract.
+
+## 7. Bundled Catalog Expectations Have Multiple Test Owners
+
+**Priority:** P3. Consolidation/removal candidate for duplicated assertions.
+
+**Relevant tests:** [internal complete catalog table](../../../internal/configuration/bundled_test.go:12),
+[public option/inventory table](../../../bundled_openrouter_contract_test.go:54),
+[HTTP payload matrix](../../../bundled_execution_contract_test.go:51), and
+[bundled smoke defaults](../../../bundled_smoke_contract_test.go:16).
+
+**Deficiency:** The current count of 17, complete ID inventory, selected
+ID-to-model mappings, tier-plugin data, and diagnostic default model recur
+across configuration, facade, wire, and diagnostic tests. A catalog update can
+require expectation edits in several layers even when loading, option handling,
+routing, and serialization are unchanged. The internal plugin assertion also
+searches `fmt.Sprint` output instead of comparing the plugin's structure.
+
+**Potential improvement:** Choose one owner for the complete maintained asset
+inventory and expected model/plugin/default data. Remove duplicate full-catalog
+counts/lists from option permutations and keep those tests focused on enable,
+disable, last-option selection, isolation, and override behavior. Retain a small
+assembled wire/diagnostic set for distinct integration risks; controlled
+overrides can isolate routing assertions from a changing catalog default.
+Use semantic structural assertions for plugin data. Keep independent expected
+asset values in the selected owner rather than deriving every expectation
+from the production loader, which could conceal an omitted or incorrect asset.
+This is consolidation, not a proposal to stop checking bundled configuration.
+
+## Findings Not Supported And Work To Preserve
+
+- No general test-flakiness, default-suite live-provider dependency, or race
+  failure was reproduced. Bounded timeout guards and channel synchronization
+  are not classified as flaky merely because a test involves concurrency.
+- The public API/type-boundary checks and their checker fixtures protect an
+  explicit architectural rule; they are not automatically obsolete because
+  they inspect structure. Stable JSON shape tests protect selected contracts.
+- Fakes at model/HTTP boundaries, counted paid-effect calls, and no-work-after-
+  rejection checks are appropriate. A fake is not deficient merely because it
+  does not model unrelated provider behavior.
+- Exhaustive domain-rule tables plus representative boundary wiring tests are
+  generally sound. No wholesale deletion of error, cancellation, capacity,
+  ownership, or prepared-lifecycle coverage is recommended.
+- Low coverage of thin wrappers, example `main` functions, defensive impossible
+  branches, and unexercised reflection variants is not independently a finding.
+  Interface-only packages do not need artificial tests to obtain a package
+  percentage. Conservative requirement inspection is not obliged to fully
+  evaluate unsupported template expressions just to cover every AST branch.
+- The standalone release-wiring fixture remains scoped to local disposable
+  repositories under the documented testing model. This audit does not propose
+  moving it into every normal test run or retesting ReleaseKit's algorithms.
+
+Prioritize the missing public privacy/source-option protection and regressions
+for the already identified defects. Make the two assertion-cleanup changes as
+focused rewrites, retaining their useful behavioral coverage. None of these
+recommendations requires new public guarantees or a coverage-percentage gate.

docs/roadmap/audit/09-interfaces.md

@@ -0,0 +1,199 @@
+# API And Interface Contract Audit
+
+Reviewed 2026-11-04 against commit `31f1e6a` and the current worktree. Existing
+audit-document moves were preserved. No production code or repository tests
+were changed.
+
+## Scope And Assessment
+
+The public root package remains a facade over configuration, execution,
+generation, capacity, rendering, validation, and observation components.
+ConfigKit owns composition before the final configuration handoff; PromptKit
+validates that handoff and captures its runtime indexes and templates. Public
+adapters translate owned request/result values and errors. Namespace views
+share execution ownership and capacity, while separately constructed engines
+remain independent.
+
+The review compared exported GoDoc, configuration and integration contracts,
+internal collaborator interfaces, call paths, and existing behavioral tests.
+It covered preparation/preflight/execution parity, prepared-handle lifecycle,
+optional values and numeric presence, configuration inspection and listing,
+injected collaborator ownership, output repair, diagnostics, observations,
+serialization, cancellation, and public error categories. It did not treat
+different intentionally documented workflows as defects or infer additional
+admission, filesystem, provider, or semantic-validation guarantees.
+
+Four findings warrant remediation. The configuration behaviors in findings 2
+and 3 are consistent with the current format reference and implementation;
+their older contradictory descriptions should be corrected rather than used
+to change those behaviors.
+
+## 1. ArtifactReader Permits Buffer Reuse Before PromptKit Can Copy It
+
+**Priority:** P2. A documented ownership permission permits data races and
+inconsistent input snapshots.
+
+**Relevant interfaces/code:** [ArtifactReader GoDoc](../../../types.go:294),
+[public reader adapter](../../../artifact_reader.go:19), and
+[copyBytes](../../../convert.go:490).
+
+**Contract defect:** The interface says that `Read` may be called concurrently
+and that the reader may reuse or mutate the returned artifact and body after
+`Read` returns. The adapter can only copy those values after that return.
+There is no synchronization or acknowledgement between return and completion
+of the copy. “Immediately copies” does not make concurrent reuse safe.
+
+**Circumstances and consequence:** A reader serializes its own `Read` calls
+with a mutex and returns a reusable buffer. Another invocation can overwrite
+that buffer as soon as the first call returns, while the first adapter is still
+copying it. Both calls follow the published reuse permission, yet preparation
+can race, mix contents, or associate metadata with the wrong bytes. This is
+not a demonstrated defect in the default file reader or in readers that
+return independent snapshots.
+
+**Evidence:** A temporary, offline public-API diagnostic used two concurrent
+`Prepare` calls and a mutex-protected reader returning a shared 1 MiB buffer.
+`go run -race` reported a write in the next `Read` racing with `copyBytes`
+through `publicArtifactReaderAdapter.Read`. The adapter's existing copy test
+establishes detachment after copying, not safety during that copy.
+
+**Potential remediation:** Resolve the return-value ownership contract before
+v1.0. The straightforward approach is for a successful `Read` to return a
+per-invocation snapshot that remains stable while PromptKit consumes it, with
+reuse permitted after the enclosing preparation/execution method returns.
+Returning fresh owned data is the simplest implementation for concurrent
+readers. If reuse immediately after `Read` itself returns must remain
+supported, it requires an explicit borrowing/release arrangement; another
+unsynchronized copy cannot provide that permission. Add focused coverage for
+the selected lifetime, without globally serializing unrelated engine work.
+
+## 2. Profile Inspection Misstates Required-Credential Semantics
+
+**Priority:** P2. The public inspection contract describes valid results as
+impossible and incorrectly limits the sources that can satisfy a requirement.
+
+**Relevant interfaces/code:** [Engine.InspectProfile](../../../engine.go:313),
+[ProfileInspection.APIKeyRequired](../../../types.go:372),
+[profile-inspection consumer example](../../consumers/pkg-promptkit.md#inspect-a-profile-before-prompt-work),
+[effective-target construction](../../../internal/configuration/targets.go:99),
+and [credential availability check](../../../internal/usecase/runner.go:611).
+
+**Contract defect:** Both method and field GoDoc claim that `APIKeyRequired`
+is mutually exclusive with a nonblank `EffectiveModelParams.APIKeyEnv`. They
+also describe the flag as requiring a direct credential or a request-level
+environment override. In fact, `api_key_required: true` suppresses a
+backend-derived environment source only when no explicit profile source was
+provided. An explicit profile environment source remains usable, as the
+[credential format reference](../../formats.md#credentials) already states.
+Inspection can therefore legitimately return both the flag and a nonblank
+environment name. The consumer example also calls every reported environment
+source optional, regardless of the flag.
+
+**Circumstances and consequence:** A profile sets `api_key_required: true` and
+`api_key_env: PROMPTKIT_INTERFACE_AUDIT_KEY`. Inspection returns `true` and
+that name; preparation succeeds using its populated environment variable
+without a direct key or request override. Consumers following the GoDoc may
+reject this valid combination or report credential readiness incorrectly.
+
+**Evidence:** A temporary public-API diagnostic demonstrated that combination
+and successful preparation. The existing
+[effective-target test](../../../internal/configuration/targets_test.go:88)
+also includes a required profile with an explicit environment source.
+
+**Potential remediation:** Describe the flag as a local credential-availability
+requirement, distinguish source selection from required availability, and
+remove the mutual-exclusion/direct-only claims. Align the consumer explanation
+with the format reference and retain inspection's intentional lack of an
+environment-value lookup. Do not reject the valid configuration or change
+the public target's stable JSON shape to make the old wording true.
+
+## 3. The Wire Contract Incorrectly Says Configured Provider Maps Replace
+
+**Priority:** P2. Conflicting published contracts predict different outbound
+provider parameters.
+
+**Relevant interfaces/code:** [integration request-body contract](../../integrations/openai-compatible-chat.md#request-body),
+[profile defaulting reference](../../formats.md#defaults-and-overrides),
+[overlayProviderMap](../../../internal/configuration/targets.go:133), and
+[runtime override merge](../../../internal/usecase/runner.go:562).
+
+**Contract defect:** The integration document says backend, profile, and request
+extra-parameter maps all use whole-map replacement. Actual configured profile
+maps recursively overlay backend defaults, including nested maps. A nonempty
+runtime `ExecutionTargetOverride.ExtraParams` instead replaces the entire
+configured map. The format reference and runtime-override GoDoc describe
+these distinct behaviors correctly.
+
+**Circumstances and consequence:** A backend supplies `backend_only: true` and
+`nested: {backend: true}`; its profile supplies `profile_only: true` and
+`nested: {profile: true}`. All entries survive into the effective target and
+built-in HTTP body. A reader of the integration contract would expect the
+backend entries to disappear, potentially leading to surprising provider
+options or ineffective attempts to clear defaults. A runtime map containing
+only `request_only: true` does remove the configured entries.
+
+**Evidence:** A temporary public-API diagnostic observed both forms through an
+injected client and confirmed the configured merged fields through a controlled
+built-in HTTP transport. No external request was made. The existing
+[effective-target test](../../../internal/configuration/targets_test.go:70)
+also exercises recursive configured overlay.
+
+**Potential remediation:** Replace the stale configuration-precedence claim
+with a short link to the format reference. Keep the integration contract
+focused on flattening the resulting effective map into the HTTP body and on
+reserved request fields. Preserve the distinct runtime replacement behavior;
+do not add another merge policy or duplicate its full configuration reference.
+
+## 4. Identity GoDoc Still Describes Removed Normalization And Version Fields
+
+**Priority:** P3. Small but concrete public-contract inconsistencies.
+
+**Relevant interfaces/code:** [ProfileInspection.ProfileID](../../../types.go:364),
+[ListPrompts GoDoc](../../../listing.go:90), and
+[literal profile identity test](../../../profile_inspection_contract_test.go:13).
+
+**Contract defect:** `ProfileInspection.ProfileID` is described as trimmed,
+but configured IDs and lookups preserve meaningful whitespace exactly.
+`ListPrompts` promises ownership of each `Version` pointer even though
+`PromptIdentity` contains only `ID`; collection-level `ReleaseVersion` belongs
+to the separate collection inventory.
+
+**Circumstances and consequence:** Inspecting the valid profile `" padded "`
+returns that exact ID, while looking up `"padded"` fails. Callers expecting the
+documented trimmed identity can fail later lookups. The nonexistent version
+field creates an erroneous expectation about available prompt metadata.
+
+**Evidence:** The existing literal-identity test explicitly protects preserved
+whitespace and failure of the trimmed lookup; a temporary public-API diagnostic
+confirmed it. The current `PromptIdentity` declaration and facade conversion
+have no version field.
+
+**Potential remediation:** State that the profile ID is the exact selected
+identity after any namespace qualification, and remove the nonexistent version
+pointer language. Keep literal identity behavior and collection metadata
+unchanged; no new normalization, version field, or compatibility guarantee is
+needed.
+
+## Existing Findings And Validation
+
+The earlier error-handling audit already records loss of owning public error
+categories for nested injected causes, loss of successful-body I/O causes, and
+pre-dispatch smoke failures incorrectly becoming concluded reports. Those are
+also interface-contract concerns, not new duplicate findings here. The hidden
+`ValidationPreparer` requirement behind the runner's `Validator` parameter is
+already recorded in the [complexity audit](03-complexity.md). No additional
+independent defect was established in the reviewed prepared-handle lifecycle,
+optional-value handling, inspection ownership, or observation/result mapping.
+
+Focused existing tests passed with the local Go 1.27.0 toolchain:
+
+```sh
+go test -race -count=1 -run 'Test(InspectProfile|ListIdentities|ScopedIdentityListings|PublicArtifactReaderAdapter|PublicLLMClientAdapter|ConfigurationInspection|LoadBuildsEffectiveTargetsAfterInheritance)' . ./internal/configuration
+```
+
+The separate temporary reader-reuse diagnostic intentionally produced a race
+report and exited unsuccessfully, substantiating finding 1 despite the green
+existing tests. Temporary programs and outputs remain outside the repository.
+All diagnostics were offline with synthetic configuration and credentials;
+no paid provider calls, new repository tests, or production changes were used.
+These checks are review evidence, not additional release gates.

docs/roadmap/audit/10-equivalence.md

@@ -0,0 +1,249 @@
+# Behavioral Equivalence And Shared Abstraction Audit
+
+Reviewed 2026-11-04 against commit `5177db0` and the current worktree. Existing
+audit-document moves were preserved. No production code or repository tests
+were changed.
+
+## Assessment And Method
+
+The public package adapts consumer values into internal configuration and
+execution components. ConfigKit owns source composition; PromptKit interprets
+its final handoff and captures indexes and templates. The runner shares
+preparation and execution machinery across public workflows, generation owns
+attempts and retries, and capacity owns admission and permits. Domain rules,
+validation, JSON-value ownership, and observation have existing shared owners.
+The architecture does not need another public abstraction or a general-purpose
+utility package to address the opportunities below.
+
+This review started from tests: direct cross-path comparisons, common test
+matrices, and separately asserted matching behavior. It then examined the
+corresponding implementation and callers to distinguish a shared responsibility
+from similar-looking code. Matching syntax, similar coverage, or duplicated
+test setup alone did not establish an extraction opportunity.
+
+Three bounded refactoring opportunities warrant consideration. They provide
+test-backed scope for findings 1, 4, and 3 respectively in the
+[complexity audit](03-complexity.md), rather than three additional independent
+bugs or duplicate remediation projects. None is a newly demonstrated behavioral
+failure or an automatic release blocker. Existing behavior, not a stronger
+equivalence promise, is the preservation target.
+
+## 1. Share The Schema Compilation Lifecycle Across Source And Execution Paths
+
+### Evidence
+
+- [TestJSONSchemaNumericConstraintsRetainJSONPrecision](../../../internal/validate/standard_validator_test.go:721)
+  runs the same exact-number constraints through operating-system and `fs.FS`
+  validators using `jsonSchemaValidatorSources`. Both must distinguish adjacent
+  large integers, preserve precise decimal boundaries, and accept the same
+  valid instances.
+- [TestJSONSchemaDecodingRequiresOneCompleteDocument](../../../internal/validate/standard_validator_test.go:805)
+  uses that same source matrix to require equivalent schema/instance framing
+  and the distinction between operational schema errors and rejected output.
+- [TestSchemaReferencesPreserveEscapedFilenames](../../../internal/validate/standard_validator_test.go:492)
+  applies the same reference cases to both source implementations, with an
+  explicit platform exception for filenames illegal on Windows.
+- [TestStandardValidatorPreparedSchemaSurvivesSourceRemoval](../../../internal/validate/standard_validator_test.go:228)
+  and [TestFSValidatorPreparedSchemaSurvivesSourceMutation](../../../internal/validate/standard_validator_test.go:416)
+  separately require the same captured schema to accept an integer-valued
+  object and reject a string-valued one after its source changes. Their source
+  operations differ, but the prepared validation responsibility matches.
+
+These tests support common schema semantics and compilation ownership; they do
+not require identical filesystem representations or diagnostic strings.
+
+### Implementation
+
+[StandardValidator and FSValidator](../../../internal/validate/standard_validator.go:74)
+each implement `PrepareValidation` and direct `validateJSONSchema`. The four
+paths repeat compiler setup, root-resource handling, cancellation boundaries,
+compilation/error wrapping, and construction or use of executable schema
+state. The direct host-file path lets its loader fetch the root, whereas the
+other paths explicitly load/register it. Decoding, compiler construction,
+schema execution, and content-result classification already have shared helpers;
+the remaining duplication is the lifecycle around them.
+
+### Opportunity And Potential Remediation
+
+Keep source-specific resolution/loading adapters, but introduce one private
+compilation operation in `internal/validate` that produces the root document
+and compiled executor for the caller's selected source. Reuse it from prepared
+validation and from the direct schema-validation callback. Retain the already
+shared artifact-validation/result logic. This makes future dialect, framing,
+number-preservation, and cancellation changes less likely to land in only some
+of the four paths.
+
+The existing package is the appropriate owner; neither a public compiler API
+nor a generic filesystem framework is needed. The related runner dependency
+clarification remains part of complexity-audit finding 1.
+
+### Tradeoffs And Test Independence
+
+Direct validation currently rejects malformed output JSON before accessing its
+schema. Eagerly implementing it as `PrepareValidation` followed by plan
+validation would change I/O and error ordering. Share compilation when that
+work is reached, not necessarily the entire entry point. Preserve contained
+host paths and symlink handling, logical `fs.FS` paths, escaped resource URLs,
+single-file sources, synchronous cancellation precedence, and operation-local
+prepared freezing. These source differences belong in their adapters.
+
+Keep the source matrix and independently specified valid/invalid instances.
+Do not compute expected results using the new compilation helper or delete
+source-specific containment/cancellation tests merely because compilation is
+shared. Provider-facing packaging of external schema references is a separate
+correctness finding, not solved by this extraction.
+
+## 2. Centralize Structural Copying Of Already-Normalized JSON Values
+
+### Evidence
+
+- [TestPublicLLMClientAdapterGivesClientOwnedNestedValues](../../../llm_adapter_internal_test.go:14)
+  requires nested provider maps, schema values, messages, and cache metadata to
+  remain independent when an injected client mutates and retains its request.
+- [TestRunnerPreparedExecutionDetachesTransferredSnapshotSources](../../../internal/usecase/prepared_execution_test.go:634)
+  separately requires nested provider maps and schema values to survive source
+  mutation, repeated `Details` calls, and execution without aliasing.
+- [TestCopyHandlesIndirectionAndIsolatesNestedMutations](../../../internal/jsonvalue/jsonvalue_test.go:156)
+  exercises the underlying JSON-value ownership responsibility with typed
+  collections and repeated acyclic children. The adjacent scalar/collection
+  tests protect compatible concrete types and exact `json.Number` values.
+- [TestCopyAndCopyMapApplyDistinctEmptyKeyRules](../../../internal/jsonvalue/jsonvalue_test.go:204)
+  explicitly demonstrates a necessary policy distinction: general schema
+  objects can have empty keys, while provider maps cannot.
+
+The common requirement is an independently owned copy of accepted nested
+values. These tests do not establish universal equivalence of validation,
+normalization, and copying on arbitrary Go values.
+
+### Implementation
+
+The facade's [copyAny/copyReflectValue](../../../convert.go:406) implements a
+separate recursive reflection copier for values projected into client requests,
+prepared details, results, and profile inspections. Internal prepared-state
+cloning uses [clonePreparedRun](../../../internal/usecase/prepared_execution.go:226)
+and [jsonvalue.Copy/CopyMap](../../../internal/jsonvalue/jsonvalue.go:30).
+The latter package already owns compatible JSON-tree copying and validation.
+
+Ordinary runtime provider values reach facade projection after JSON-value
+validation/copying. The runner also copies its schema document through
+[structuredOutputFromValidationPlan](../../../internal/usecase/runner.go:419)
+before projecting it. Thus these normal runtime paths have a common accepted
+value domain, even though their entry-point policies are different.
+
+### Opportunity And Potential Remediation
+
+Move ownership of trusted JSON-tree cloning into `internal/jsonvalue`, reusing
+its compatible structural-copy machinery where semantics match, so the facade
+does not maintain another general reflection traversal. Keep public/domain DTO
+field projection in the facade. A small internal trusted-value clone operation
+is a possible approach; reuse of existing fallible operations is another where
+the caller can preserve the existing error boundary. Select the simpler option
+after inventorying the concrete callers and values, rather than introducing a
+configurable reflection framework.
+
+The meaningful payoff is removing an independent traversal implementation,
+not wrapping the existing facade copier under another name. This does not
+require consolidating every pointer or slice helper in the repository.
+
+### Tradeoffs And Test Independence
+
+`Copy` validates and bounds values and normalizes null/indirection; the facade
+copier assumes prior acceptance and can preserve typed nils and pointers.
+Neither is a drop-in replacement for the other on all possible inputs.
+Preserve observed types, nil behavior, exact numbers, schema empty keys, and
+caller ownership at the relevant boundary. An infallible projection must not
+silently turn a new copy error into a zero value, a panic, or an additional
+consumer restriction.
+
+The [configuration-boundary ADR](../../adr/0001-configuration-boundary.md)
+also remains authoritative. `providerValueForMapping` serves a different,
+pre-handoff source-encoding purpose and defers unsupported-value errors to
+ConfigKit. It must not be routed through a rejecting runtime copier as part of
+this work. Source number conversion is separately recorded in the correctness
+audit; visual similarity between its reflection switch and a copier is not
+enough to equate their contracts.
+
+Keep literal expected values and mutation-based assertions in both facade and
+internal tests. A shared production copier must not also manufacture their
+expected snapshots. Retain focused tests for policy differences rather than
+replacing them with one “all copy functions are equivalent” test.
+
+## 3. Share If/With Requirement-Branch Orchestration, Not Template Execution
+
+### Evidence
+
+- [TestAnalyzeRequirementsClassifiesKnownConditions](../../../internal/prompt/requirements_test.go:268)
+  verifies required/optional classification for selected and skipped `if`
+  branches, and directly compares the recognized branch choice with the real
+  Go renderer. This anchors recognized condition semantics in actual rendering.
+- [TestAnalyzeRequirementsHandlesWithDotAndNestedReachability](../../../internal/prompt/requirements_test.go:332)
+  requires the same selected/skipped classification for `with` and nested
+  `if`, while separately protecting `with`'s changed dot and preserved root
+  caller-map access.
+- [TestAnalyzeRequirementsReportsUnresolvedDecisionAndContentConditions](../../../internal/prompt/requirements_test.go:369)
+  protects conditional dependencies and meaningful incompleteness for unknown
+  decisions. It also establishes that an unresolved condition around plain
+  text alone does not make discovery incomplete.
+- [TestAnalyzeRequirementsDoesNotHideCallerMapScopes](../../../internal/prompt/requirements_test.go:14)
+  compares several scope forms against actual template execution while allowing
+  incomplete analysis for constructs outside recognition scope.
+
+The tests support one reachability/evidence policy for recognizable branches,
+with explicit scope differences. They do not require the inspector to execute
+or completely interpret arbitrary templates.
+
+### Implementation
+
+The `IfNode` and `WithNode` cases in
+[requirementWalker.visit](../../../internal/prompt/requirements.go:451)
+independently perform inactive-path traversal, condition evaluation, then/else
+reachability selection, dependency-counter comparison, and limitation recording.
+`with` adds a different body-dot calculation. `branchReach`, condition
+evaluation, and `template.IsTrue` already provide shared lower-level behavior,
+but their surrounding branch protocol is repeated.
+
+### Opportunity And Potential Remediation
+
+Extract a small package-private branch traversal operation that owns the common
+reachability and evidence sequence, with explicit body/else scope handling.
+Leave node recognition and `withDotIdentity` visible at the node-specific
+boundary. Keeping this in `internal/prompt` would allow a change to skipped-path
+handling or incompleteness reporting to be reviewed and implemented once.
+
+This is the narrowly supported part of complexity-audit finding 3. It does not
+justify sharing the requirement walker with the real template executor or
+building an interpreter abstraction.
+
+### Tradeoffs And Test Independence
+
+An abstraction that treats `if` and `with` identically would lose caller-map
+scope information. Preserve `with`'s body-dot behavior, else scope, inactive
+computed-name handling, and the placement of dependency bookkeeping used to
+decide whether an unresolved condition matters. A helper with numerous mode
+flags or callbacks would obscure those differences; prefer a small explicit
+operation over a generalized AST visitor framework.
+
+Retain separate node-specific assertions and the independent standard-library
+rendering comparisons. Extracting production traversal is not a reason to
+calculate expected inventories using the production branch helper or to
+strengthen `Complete` into an execution-readiness guarantee.
+
+## Recommendation And Validation
+
+Start with schema compilation, which has the clearest shared lifecycle and
+cross-source test matrix. The branch helper is a smaller local follow-up.
+Treat JSON-copy consolidation as a more careful ownership refactor: remove a
+traversal only after its caller/policy differences are accounted for. No new
+package boundary is necessary solely to carry out these changes, and retaining
+small source/boundary-specific adapters is appropriate.
+
+Existing offline tests passed with the local Go 1.27.0 toolchain:
+
+```sh
+go test -race -count=1 ./internal/validate ./internal/prompt ./internal/jsonvalue
+go test -race -count=1 -run 'Test(PublicLLMClientAdapter|PreparedExecutionFreezesSourcesAndReturnsIndependentDetails|RunnerPreparedExecutionDetachesTransferredSnapshotSources|RunnerRunAndPrepareResolveSameProfileAndEffectiveSettings|RunnerPreparationNormalizesOutputContractConsistently|ValidateRequestMatchesPrepareContract)' . ./internal/usecase
+```
+
+Passing tests substantiate the reviewed current behavior, not equivalence of a
+future refactor. No new tests, production changes, live provider requests,
+benchmarks, or additional release gates were introduced by this audit.

docs/roadmap/audit/11-boundaries.md

@@ -0,0 +1,229 @@
+# Package Cohesion And Architectural Boundary Audit
+
+Reviewed 2026-11-04 against commit `b39460a` and the current worktree. Existing
+audit-document moves were preserved. No production code, tests, dependencies,
+or functionality were changed.
+
+## Scope, Method, And Architectural Assessment
+
+Reviewed the 17 repository-owned internal Go packages, their production import
+relationships, engine assembly, relevant callers, and tests. The public facade
+was checked where it establishes the boundary around those packages. Third-party
+package implementations were outside the audit's scope.
+
+The [current component inventory](../../internal/overview.md) broadly matches
+the implementation. Configuration owns the final ConfigKit handoff and captured
+resource indexes; the engine assembles collaborators; use cases coordinate
+selection, preparation, execution, and diagnostics. Generation owns dispatch
+and retries, capacity owns admission and permits, and outbound transport,
+rendering, artifacts, schema validation, observations, and shared values have
+identifiable owners. No repository-wide package reorganization is warranted.
+
+The public facade does not leak internal types or receive reverse production
+imports from internal packages in the checks run below. Deliberate boundaries
+remain useful: configuration completion differs from request overrides;
+diagnostic orchestration differs from ordinary execution; domain invariants
+differ from source-specific presence and error handling. A broad package, a
+small interface-only package, or an import of another internal component is not
+itself a cohesion defect.
+
+Three focused ownership improvements warrant consideration. They are
+maintainability opportunities, not newly demonstrated runtime bugs or automatic
+release blockers.
+
+## 1. Retire The Test-Only Capacity Client As A Production Scheduling Surface
+
+**Relevant code:** [capacity.NewClient and client.Generate](../../../internal/capacity/client.go:17),
+[generation constructors and dispatch](../../../internal/generation/coordinator.go:37),
+[engine assembly](../../../internal/engine/engine.go:157), and
+[runner scheduling test](../../../internal/usecase/runner_test.go:2391).
+
+**Basis:** The engine selects `generation.NewBuiltIn` or `NewInjected` for every
+production engine. The coordinator owns active-permit acquisition around each
+attempt, retry waits, observation, and diagnostic evidence. The earlier
+`capacity.NewClient` still provides a separate model-client wrapper that
+acquires/releases a permit around an opaque call. Both use the same manager;
+there are not two independent FIFO algorithms or pools.
+
+Caller analysis found only tests invoking the older constructor: capacity tests
+and a runner test that assembles initial/repair scheduling with this wrapper.
+The [capacity document](../../internal/capacity.md#fifo-generation-permits)
+acknowledges its retention for focused callers/tests, but no production caller
+was found. Its presence leaves capacity owning a model-client adapter after
+production dispatch responsibility moved to generation, and makes one runner
+assembly test exercise a different scheduling composition than the engine.
+
+**Consequences:** The extra adapter and its documented role remain maintenance
+surface without a production need. Future scheduling changes can require
+reasoning about a legacy assembly alongside the real one. Green tests of the
+wrapper alone do not demonstrate that coordinator dispatch applies the same
+permit lifetime. Existing coordinator and facade tests do protect that actual
+path, so this is not a claim that production scheduling is untested or broken.
+
+**Potential remediation:** Keep the manager and its admission/FIFO primitives
+in `internal/capacity`, and generation dispatch integration in
+`internal/generation`. Replace the runner test's wrapper with the production
+injected coordinator. For focused capacity scenarios, test `Manager.Acquire`
+directly or retain the small one-shot adapter as test-owned scaffolding rather
+than a production constructor. Once callers are adjusted, the production
+wrapper can be retired and its documentation updated.
+
+Preserve the useful FIFO, cancellation-race, independent-pool, and panic-release
+assertions when adjusting test ownership. This finding does not call for moving
+queue internals into generation, deleting those scenarios, or wrapping the
+coordinator in another capacity client, which could acquire the same limit
+twice. Retaining the wrapper is viable if a real internal production need is
+identified; current test reachability alone does not establish one.
+
+## 2. Put OpenRouter Authentication Protocol Details Behind The Outbound Adapter
+
+**Relevant code:** [authenticationProbe.runOpenRouterKey](../../../internal/usecase/smoke_authentication.go:66),
+[recognizedOpenRouterKeyResponse](../../../internal/usecase/smoke_authentication.go:139),
+[diagnostic endpoint derivation](../../../internal/usecase/smoke_selection.go:166),
+and [HTTP authentication transport](../../../internal/llm/http_evidence.go:111).
+
+**Basis:** `internal/usecase` currently interprets OpenRouter-specific HTTP
+statuses and the account-response keys `data`, `label`, `is_free_tier`, and
+`usage`. It owns JSON envelope parsing, identifying-field type checks, and body
+consumption. The same package derives the OpenRouter-relative `/key` endpoint.
+Meanwhile `internal/llm` owns authenticated HTTP execution, credential
+resolution, evidence collection, and response ownership mechanics. One provider
+integration is therefore divided between the application coordinator and its
+outbound adapter.
+
+This division is different from a use case consuming factual HTTP evidence:
+the precise external account document and provider endpoint are protocol
+knowledge. The [architecture policy](../../policy/architecture.md#architectural-approach)
+calls for encapsulating external integration details in internal adapters.
+The parser is already functionally isolated, making the ownership seam fairly
+small. The authentication tests in
+[smoke_authentication_test.go](../../../internal/usecase/smoke_authentication_test.go:64)
+also show that this is provider protocol behavior rather than application
+selection policy.
+
+**Consequences:** A change to OpenRouter's account endpoint or identifying
+response fields requires changing application-layer code as well as the
+transport integration. Adding another authentication adapter would extend
+provider-specific branching inside the use-case package. This complicates the
+otherwise clear distinction between choosing a diagnostic and implementing
+the upstream protocol.
+
+**Potential remediation:** Move the OpenRouter request/response adapter,
+including its default destination and response recognition, into
+`internal/llm` alongside the existing diagnostic transport. Have the use case
+consume a small factual outcome/evidence value. Keep target selection,
+configured applicability, unsupported-operation reporting, caller cancellation,
+overall completion, and report assembly in the use case. The existing
+consumer-owned diagnostic interface is a useful seam; engine assembly can
+supply the concrete outbound implementation without requiring protocol parsing
+inside the coordinator.
+
+An alternative is a focused internal diagnostic-adapter package if separation
+from inference transport becomes useful. Another package or a provider plugin
+framework is not necessary for the current single protocol adapter. Do not
+introduce a new dependency from the outbound adapter back into `usecase`, or
+move application report/completion policy into transport merely to relocate
+the parser. The diagnostic feature legitimately deals with HTTP facts; a
+blanket prohibition on HTTP types or status observations in use cases is not
+implied by this finding.
+
+Preserve anonymous authentication probing, explicit endpoint overrides,
+neutral tri-state facts, private body/credential handling, and the absence of
+ordinary observer events. The previously recorded unbounded authentication-body
+read is a separate resource/security finding; relocation alone does not fix it.
+
+## 3. Narrow Smoke-Value Copy Helpers To Their Actual Ownership Boundaries
+
+**Relevant code:** [domain.CloneSmokeTestRequest](../../../internal/domain/smoke_diagnostics.go:157),
+[domain.CloneSmokeTestReport](../../../internal/domain/smoke_diagnostics.go:179),
+[smokeReportTarget](../../../internal/usecase/smoke_tester.go:432), and
+[public smoke-value conversion](../../../smoke_diagnostics.go:279).
+
+**Basis:** `CloneSmokeTestRequest` is called only by its own domain test;
+actual request ownership is established by public-to-domain conversion.
+`CloneSmokeTestReport` has one production caller: `smokeReportTarget` creates
+an otherwise empty report containing a target, clones the full report, and
+extracts just that target. Whole-report facts, timing, and attempt cloning have
+no runtime use through that caller. Public report conversion independently
+copies the actual facts, timings, attempts, and provider map into public types.
+
+Thus the domain package maintains a broad request/report-copy API, but the
+real internal consumer needs only a detached target/provider map. The
+whole-report ownership functionality is exercised by its domain test rather
+than by actual whole-report production use. Public DTO conversion is a
+legitimate separate responsibility, not something that should disappear into
+domain types or public aliases.
+
+**Consequences:** Smoke-value changes have additional copy lists and tests to
+maintain even where those copies are not needed in production. A temporary
+report container obscures the narrow target-detachment operation. Keeping a
+general cloner solely because its own test calls it reverses the relationship
+between production responsibility and supporting tests.
+
+**Potential remediation:** Express the target-copy operation directly at its
+use-case boundary using existing JSON-value ownership, or a small target clone
+helper if actual callers justify one. Retain public request/report projection
+in the facade and shared JSON-tree copying in its existing owner. Retire the
+unused request cloner and the unnecessary whole-report clone surface after
+confirming no additional callers need them. Remove only clone-specific helper
+code that is then unused. The backend-cloning helpers have real registry
+consumers and are outside the proposed retirement; retain that metadata
+ownership behavior.
+
+Preserve provider-map error propagation, caller-owned result values, and
+credential/endpoint redaction. Keep the meaningful
+[public conversion ownership test](../../../smoke_conversion_internal_test.go:13)
+and relevant use-case report tests. If genuine internal whole-report snapshots
+are needed, retaining an explicitly used domain cloner is a reasonable
+alternative. This finding does not require sharing public/internal DTO types
+or routing public conversion through a full internal clone followed by another
+copy. Coordinate any JSON traversal consolidation with
+[audit 10](10-equivalence.md#2-centralize-structural-copying-of-already-normalized-json-values).
+
+## Related Existing Findings
+
+The runner's declared `Validator` dependency and actual `ValidationPreparer`
+requirement remain the interface-boundary clarification in
+[complexity-audit finding 1](03-complexity.md#1-consolidate-schema-compilation-and-express-the-runners-actual-interface).
+The facade's generic JSON reflection copier has an existing internal owner
+identified in audit 10. These are relevant ownership concerns, not additional
+independent findings to count again. Large orchestration functions and repeated
+schema compilation within an already coherent package likewise retain their
+existing complexity/equivalence remediation scope.
+
+## Reachability And Validation Evidence
+
+Production imports were enumerated with `go list` and compared with the graph
+and source callers. Rapid Type Analysis used the locally available pinned
+`golang.org/x/tools` v0.52.0 command, with reporting restricted to
+repository-owned internal packages:
+
+```sh
+go list -f '{{.ImportPath}}|{{join .Imports " "}}' ./internal/...
+go run golang.org/x/tools/cmd/deadcode@v0.52.0 -test -filter='^go.maximumdirect.net/promptkit/internal/' ./...
+go run golang.org/x/tools/cmd/deadcode@v0.52.0 -filter='^go.maximumdirect.net/promptkit/internal/' ./...
+```
+
+With tests included, the tool reported only the unused test helper
+`internal/usecase/runner_test.go:stringPtr`. That small test cleanup does not
+justify a separate architectural finding. Test-only production helpers remain
+reachable to this analysis when their tests call them.
+
+Without tests, the roots are the repository's example executables, not every
+possible downstream library consumer. The output includes valid public
+workflows that those examples do not exercise, so it is not a deletion list.
+The capacity-wrapper and request-cloner observations above were additionally
+checked against actual source references and assembly; graph in-degree or
+example reachability alone was not treated as proof of obsolescence. The
+analysis also covers only the selected local build configuration.
+
+The following existing offline checks passed with local Go 1.27.0:
+
+```sh
+go test -race -count=1 -run 'Test(PublicFacadeDoesNotExposeInternalTypes|InternalProductionPackagesDoNotImportFacade|RepositoryDoesNotImportFormerModule|SmokeTestConcludedInferenceMismatchHasNilGoErrorAndDetachedValues)' .
+go test -race -count=1 ./internal/capacity ./internal/generation ./internal/domain ./internal/usecase
+```
+
+No new release gate, consumer contract, provider call, production edit, or test
+removal is proposed as part of this audit pass. The tools left module metadata
+unchanged.

docs/roadmap/audit/12-deadcode.md

@@ -0,0 +1,233 @@
+# Dead Code And Obsolete Functionality Audit
+
+Reviewed 2026-11-04 against commit `3b76031` and the current worktree. Existing
+audit-document moves were preserved. No production code, tests, dependencies,
+or functionality were changed.
+
+## Scope And Assessment
+
+Reviewed repository-owned Go packages, public entry points, configuration and
+runtime values, collaborator interfaces, default/optional paths, embedded
+resources, build constraints, and the release launchers and their wiring test.
+The architecture remains a public facade over captured configuration, use-case
+orchestration, generation/capacity, and focused internal adapters. The former
+standalone configuration repositories have already been retired; this review
+found smaller residues rather than another abandoned subsystem.
+
+The review combined Rapid Type Analysis, Staticcheck unused/data-flow checks,
+graph caller discovery, source inspection, and the format, internal, policy,
+and roadmap documents. Tests count as real callers, but a helper's self-test
+does not by itself establish a production need. Conversely, absence from the
+examples or default execution path is not evidence that an optional public
+feature is obsolete.
+
+Four cleanup areas warrant consideration. These are low-priority reductions in
+state and maintenance surface, not newly demonstrated runtime bugs. None calls
+for deleting a supported public feature or imposing a new release gate.
+
+## 1. Remove Backend Presence Bookkeeping With No Runtime Consumer
+
+**Relevant code:** [loaded.backends](../../../internal/configuration/load.go:122),
+[backendFields and presence capture](../../../internal/configuration/load.go:215),
+[effective backend construction](../../../internal/configuration/targets.go:14),
+and [Backend.QueueCapacitySet](../../../internal/domain/domain.go:187).
+
+**Evidence:** Construction allocates a per-backend field-state map and fills it
+through `captureFields(..., state.backends)`. No production code reads the
+recorded backend states. The only lookup of the captured queue state is in
+[TestLoadDefaultsAbsentQueueCapacity](../../../internal/configuration/load_test.go:755).
+Unlike profile presence, backend presence is not consumed while constructing
+effective execution targets.
+
+`QueueCapacitySet` is also assigned during target construction and asserted in
+tests, but never read to select a production behavior. The current assignment
+is simply whether concurrency is positive; it no longer distinguishes an
+explicit queue setting from a derived default. The registry's capacity-policy
+projection uses `ConcurrencyLimit` and `QueueCapacity`, not this flag.
+
+**Why obsolete:** Queue defaulting and its provenance are already handled by
+[resolveBackends](../../../internal/configuration/resolve_targets.go:11) and
+ConfigKit's derived-default record. Typed final values drive runtime capacity;
+the retained side map and Boolean do not contribute to that behavior or to the
+public inspection view. No documented compatibility or planned-feature need for
+these extra records was found.
+
+**Potential remediation:** Remove the backend-only presence capture, its field
+list and retained map, and the unused runtime Boolean/assignment. Adjust tests
+to assert effective queue behavior and, where relevant, the existing ConfigKit
+inspection/provenance result. Keep `captureFields`, profile field-state
+tracking, and prompt-input metadata presence: those have actual consumers and
+preserve required absent/null/explicit-value distinctions. Retain explicit
+zero queue capacity, derived defaulting, and operator history unchanged.
+
+This removes unnecessary construction work and retained per-backend maps; it
+does not justify dropping configuration provenance or changing its ownership
+under the [configuration-boundary ADR](../../adr/0001-configuration-boundary.md).
+
+## 2. Retire Legacy Fields In The Captured Prompt Representation
+
+**Relevant code:** [PromptDefinition.OutputFormat](../../../internal/domain/domain.go:148),
+[PromptMessageTemplate.ContentFile](../../../internal/domain/domain.go:173),
+[prompt materialization](../../../internal/configuration/prompts.go:58),
+[resolveOutputContract](../../../internal/usecase/runner.go:639), and
+[definition hashing](../../../internal/promptdef/hash.go:12).
+
+**Evidence:** Supported engine construction requires and validates a complete
+output contract. Materialization sets `definition.Validation` to that contract
+and duplicates its format into `definition.OutputFormat`. The runner reads
+the duplicate only when `Validation.Format` is empty, a condition that the
+captured definitions supplied by engine construction cannot have. The standard
+runner test-definition helper fills both fields with the same value; no
+supported public repository-injection path supplies an alternative definition.
+
+For messages, materialization resolves the real source `content_file` into
+captured `Content` and does not populate the domain `ContentFile` field. The
+renderer uses `Content`/captured templates, not that path field. Its one
+nonempty assignment is a renderer fixture demonstrating that already-loaded
+content is rendered without opening a file. The source-format
+`promptMessageConfig.ContentFile` remains actively used and is a different
+field.
+
+**Why obsolete:** The runtime representation still carries source-era message
+path data and a second output-format source after configuration loading became
+the authoritative materialization boundary. The empty-format fallback is
+disconnected from supported engine definitions. No current documentation or
+roadmap records a compatibility requirement for these internal duplicate fields.
+
+**Potential remediation:** Use `PromptDefinition.Validation.Format` as the one
+captured format and remove the legacy duplicate/fallback. Remove only the
+runtime message path field, adjusting its fixture to exercise captured text.
+Preserve the actual YAML `content_file` feature, source-relative resource
+resolution, ConfigKit origin inspection, and construction-time capture. Also
+preserve the separate, supported request-override rule that an empty replacement
+`OutputContract.Format` defaults to text.
+
+**Serialization caveat:** These fields are included indirectly by
+`HashDefinition`'s `json.Marshal`, even when a message path is empty. They are
+therefore not completely inert bytes: removing them changes current prompt
+hash values. Account for that effect explicitly and protect the documented
+opaque-equality relationships between inspection and preparation. Do not claim
+bit-identical hashes after changing the hash input, or invent a permanent hash
+encoding guarantee merely to perform this cleanup. The fields are internal
+representation, not part of the public stable DTO shapes.
+
+## 3. Remove Unused Diagnostic State And A Discarded Helper Result
+
+**Relevant code:** [reachabilityProbeOutcome.initialEndpoint](../../../internal/usecase/smoke_reachability.go:24),
+[HTTP report assembly](../../../internal/usecase/smoke_tester.go:376),
+[fileSource](../../../internal/engine/engine.go:211), and
+[SetSchemaFile](../../../internal/engine/engine.go:75).
+
+**Evidence:** The reachability probe assigns its private outcome's
+`initialEndpoint`, but the smoke tester never reads that member. It records
+the real public attempt's `InitialEndpoint` from the resolved report target.
+Only `TestReachabilityProbeRemovesURLCredentials` inspects the private echo.
+Actual sanitized request/evidence and report behavior are established through
+other values.
+
+Separately, `fileSource` returns a filesystem, file name, source-directory
+string, and error. Its only caller, `SetSchemaFile`, always discards the
+directory result. The directory is still needed internally to construct
+`os.DirFS`; exposing it as another return value serves no caller.
+
+**Potential remediation:** Remove the unused probe-outcome member and its
+assignment; retain credential-removal assertions against the actual request,
+evidence, or public diagnostic report. Reduce `fileSource`'s private signature
+to the consumed results. Keep its directory calculation, file checks, selected
+filesystem/root, and the public `WithSchemaFile` option intact.
+
+Neither cleanup removes `SmokeTestAttempt.InitialEndpoint`, destination
+evidence, or single-schema-file support. Those public features are live even
+though an extra private field/result is unnecessary.
+
+## 4. Remove Orphan Helpers Whose Only Callers Are Their Own Tests
+
+**Relevant code and evidence:**
+
+- [filecatalog.RelativePath](../../../internal/filecatalog/catalog.go:71) is
+  called only by `TestRelativePathNested`. Current source discovery and
+  source-relative resource resolution use other helpers; no production or
+  public-facade caller uses this wrapper around `filepath.Rel`.
+- [resourceNamespaceFromID](../../../internal/domain/resource_identity.go:34)
+  is called only by `TestResourceNamespaceFromID`. Actual namespace inventory
+  and filtering paths do not call it. The exported identity-validation and
+  qualification operations remain active and are not candidates for removal.
+- [runner_test.go:stringPtr](../../../internal/usecase/runner_test.go:2592)
+  has no callers, including in tests. Both reachability analysis and
+  Staticcheck `U1000` report it.
+
+Caller searches found no reflection, registration, build-constrained caller,
+or supported external API exposing these helpers. The first two belong to
+internal packages and have only their direct self-tests; no documented future
+or compatibility purpose was found. The third is already test-only and unused.
+
+**Potential remediation:** Remove the orphan helpers and their now-irrelevant
+self-tests, keeping behavioral coverage of real discovery, namespace
+inventory/filtering, and identity qualification. Remove the unused pointer
+helper without replacing it. There is no reason to introduce a new abstraction
+or move the self-tests to another layer solely to preserve these unused APIs.
+
+## Intentional Retention And Related Findings
+
+The following distinctions prevent removal based on misleading reachability:
+
+- `WithSchemaFile` and `InlineWithURI` are supported public helpers despite
+  being unreachable from the current test executables. The former's missing
+  public wiring coverage is already in [audit 08](08-testing.md); it is not
+  evidence that the option should be deleted.
+- The older capacity client wrapper and no-repair runner constructor are
+  expressly retained for focused internal callers/tests in the current
+  [capacity](../../internal/capacity.md#fifo-generation-permits) and
+  [runner](../../internal/runner.md#collaborators) documentation. This audit
+  does not classify them as dead functionality or recommend deleting them.
+  [Audit 11](11-boundaries.md) discusses the wrapper's architectural placement;
+  altering that deliberate retention is a separate decision.
+- Repository capability fallbacks, standalone rendering/validation, injected
+  clients' nil usage-presence compatibility, nil-validator/admitter behavior,
+  and optional provider settings have meaningful callers or documented
+  purposes. Normal engine assembly taking another path does not obsolete them.
+- ConfigKit YAML fields and stable JSON DTO fields can be used by reflection;
+  formatter methods and interfaces can be invoked indirectly. Embedded
+  OpenRouter and smoke-schema assets, the Linux FIFO test, and release scripts
+  likewise have consumers outside a simple direct-call list. They were not
+  treated as dead. No unused public configuration option or abandoned release
+  launcher was established.
+
+The domain smoke-request/full-report cloners have a narrow-or-unused consumer
+surface already described in [audit 11, finding 3](11-boundaries.md#3-narrow-smoke-value-copy-helpers-to-their-actual-ownership-boundaries).
+That ownership/remediation scope is not counted again here. The schema and
+JSON-copy consolidations in prior audits likewise remain their existing work,
+not an excuse to delete intentional alternative implementations wholesale.
+
+## Tooling And Validation
+
+The checks used local Go 1.27.0 and pinned audit-tool versions:
+
+```sh
+go run golang.org/x/tools/cmd/deadcode@v0.52.0 -test -filter='^go.maximumdirect.net/promptkit(/|$)' ./...
+go run honnef.co/go/tools/cmd/staticcheck@v0.8.2 -checks 'U1000,SA4*' ./...
+go list -f '{{.ImportPath}} {{join .GoFiles " "}} {{join .IgnoredGoFiles " "}}' ./...
+```
+
+`deadcode -test` reported the two public helpers and unused test pointer helper
+listed above. Staticcheck reported only `stringPtr` (`U1000`) and exited with
+status 1 because of that finding; it reported no additional selected `SA4*`
+diagnostics. These are audit observations, not successful lint sign-offs.
+Neither tool establishes that all tested or serialized state is necessary,
+which is why the write-only and redundant fields required manual review.
+
+Rapid Type Analysis starts from executable roots; tests add roots, and example
+programs do not exercise every library consumer path. The analysis is local-build
+evidence, not proof about all platforms or downstream programs. In particular,
+platform-specific schema handling and optional library operations were retained.
+
+Existing offline race-enabled tests passed:
+
+```sh
+go test -race -count=1 ./internal/configuration ./internal/domain ./internal/filecatalog ./internal/prompt ./internal/engine ./internal/usecase
+go test -race -count=1 -run 'Test(ConfigurationInspection|PreparedExecution|ValidateRequestMatchesPrepareContract|AppendedMessagesComposeFrozenEffectivePrompt)' .
+```
+
+Tool execution left `go.mod` and `go.sum` unchanged. No production changes,
+test removals, provider calls, or new validation gates were made during this
+review.

docs/roadmap/audit/catalog.md

@@ -0,0 +1,570 @@
+# Consolidated Pre-1.0 Audit Catalog
+
+Consolidated on 2026-11-04 from the 11 audit documents in this directory.
+Their 41 numbered findings are represented by 34 entries below. Entries that
+describe the same underlying issue share one remediation scope; missing
+regression coverage for an identified defect is included with that defect.
+Some broad testing findings consequently support more than one entry.
+
+The order reflects consequence, exposure, and the value of clarifying public
+contracts before v1.0. Confirmed defects and conditional security exposure come
+first, followed by contract/test gaps, resource efficiency, and maintainability.
+This ordering does not upgrade the original severities or turn every cleanup
+into a release gate. The source audits retain their evidence, qualifications,
+and validation results. None of these entries is marked remediated by this
+documentation-only consolidation.
+
+## 1. Update The Audited Maintainer Toolchain For Applicable Security Fixes
+
+**Conditional security exposure.** The audited Go 1.27.0 toolchain predates
+networking fixes applicable to HTTP clients, including HTTP/2 CPU exhaustion
+and conditional certificate-trust exposure. Use a patched maintainer toolchain
+and repeat the focused security assessment against it. Actual downstream
+exposure depends on the toolchain that builds the consumer and its transport
+and trust configuration; raising PromptKit's minimum Go requirement is a
+separate compatibility decision, not an implied remedy.
+
+Sources: [Security 1](07-security.md#1-the-audited-go-toolchain-predates-applicable-networking-security-fixes).
+
+## 2. Bound Authentication-Probe Response Consumption
+
+**Confirmed resource/security defect.** Authentication diagnostics read a
+successful upstream body with unbounded `io.ReadAll`. A deadline does not bound
+the bytes a fast server can supply. Introduce finite response consumption and
+clean closure, with an explicit choice of byte budget and overflow reporting.
+Overflow should not be interpreted as proof of rejected authentication.
+Relocating the provider adapter in entry 26 would not itself fix this defect.
+
+Sources: [Resources 1](05-resources.md#1-authentication-diagnostics-read-an-unbounded-upstream-body),
+[Security 2](07-security.md#2-authentication-probes-permit-unbounded-remote-body-memory-consumption).
+
+## 3. Avoid Exponential Expansion During Convenience Source Conversion
+
+**Confirmed resource defect.** `ProfileSource` and `BackendSource` conversion
+expand repeated references before ConfigKit performs its bounded snapshot.
+A small shared acyclic graph can therefore cause exponential work and
+allocation. Preserve shared references during conversion, with suitable cycle
+and slice-view handling, rather than duplicating ConfigKit's validation or
+introducing PromptKit-side raw-source limits. Add focused shared-graph and
+deferred-failure regressions for both convenience helpers.
+
+Sources: [Resources 2](05-resources.md#2-convenience-source-conversion-expands-shared-graphs-before-bounded-loading),
+[Testing 3](08-testing.md#3-programmatic-provider-value-conversion-lacks-representative-value-cases).
+
+## 4. Resolve The ArtifactReader Buffer-Ownership Contract
+
+**Confirmed race under the documented collaborator contract.** The public
+reader contract permits buffer reuse after `Read` returns, before PromptKit
+can finish copying it. Concurrent preparation with a compliant custom reader
+can race. Clarify ownership/lifetime per invocation and align adapters, GoDoc,
+and concurrency tests. This needs a deliberate public-contract decision before
+v1.0: an immediate copy alone cannot guarantee safety if the reader is allowed
+to mutate the buffer immediately. No default-reader defect was established.
+
+Sources: [Interfaces 1](09-interfaces.md#1-artifactreader-permits-buffer-reuse-before-promptkit-can-copy-it).
+
+## 5. Preserve Exact Numbers In Programmatic Provider Sources
+
+**Confirmed data-integrity defect.** Convenience source conversion falls back
+from integer parsing to `float64` without checking precision, silently changing
+some accepted `json.Number` values. Preserve an exact ConfigKit-compatible
+representation, or defer an unsupported-value error to ConfigKit instead of
+rounding or converting the number into a string. Regression coverage should
+include exact-number boundaries, nested typed collections, mutation isolation,
+and deferred unsupported/cyclic-value errors for both source helpers.
+
+Sources: [Correctness 1](01-correctness.md#1-provider-source-helpers-silently-change-exact-json-numbers),
+[Testing 3](08-testing.md#3-programmatic-provider-value-conversion-lacks-representative-value-cases).
+
+## 6. Make Provider-Native Schema Constraints Self-Contained
+
+**Confirmed implementation gap; packaging policy needs selection.** Local
+validation freezes referenced schema resources, but the provider receives only
+the root document and cannot resolve its local external references. Select
+correct resource packaging, or an explicit local-only fallback policy, without
+naively inlining recursive schemas or silently dropping native constraints.
+Add a wire-level regression demonstrating a usable provider constraint, not
+merely successful local compilation. Entry 21 is a separate compilation
+refactor and does not solve this packaging problem.
+
+Sources: [Correctness 2](01-correctness.md#2-external-schema-resources-do-not-reach-the-provider-constraint),
+[Testing 5](08-testing.md#5-multi-resource-schema-tests-do-not-establish-a-usable-provider-constraint).
+
+## 7. Preserve The Owning Operation's Public Error Category
+
+**Confirmed classification defect.** Public error conversion searches nested
+causes in a global priority order, allowing a collaborator's sentinel to hide
+the enclosing operation's category. Map the owning boundary independently
+while retaining the original cause for `errors.Is`/`errors.As`. Add composed
+cause regressions; simply reordering global checks can break legitimate direct
+not-found versus load-failure distinctions.
+
+Sources: [Error handling 1](06-error-handling.md#1-nested-collaborator-errors-can-hide-the-owning-operations-public-category),
+[Testing 4](08-testing.md#4-error-tests-miss-composed-causes-and-pre-dispatch-diagnostic-failures).
+
+## 8. Distinguish Response Stream Failures From Malformed JSON
+
+**Confirmed error-handling defect.** Non-context I/O failures while decoding a
+successful HTTP response or checking its end are reported as malformed JSON,
+losing the transport cause; complete JSON followed by a read failure can also
+be misreported as trailing data. Retain and classify reader failures separately
+using the existing redacted transport boundary. Preserve cancellation
+precedence and genuine malformed/truncated/oversized-content handling. Cover
+both read locations without adding transport retries.
+
+Sources: [Error handling 2](06-error-handling.md#2-successful-status-body-read-failures-are-misreported-as-malformed-json),
+[Testing 4](08-testing.md#4-error-tests-miss-composed-causes-and-pre-dispatch-diagnostic-failures).
+
+## 9. Return Invalid Diagnostic Request Construction As A Go Error
+
+**Confirmed diagnostic-contract defect.** Some invalid inference settings fail
+before dispatch yet produce a completed report with a nil Go error. Propagate
+local request-construction failures through `ErrInvalidRequest` and add a
+no-dispatch regression. Preserve intentionally completed diagnostic outcomes
+for missing credentials, capacity outcomes, and unsupported or inapplicable
+operations; not every no-dispatch outcome is invalid configuration.
+
+Sources: [Error handling 3](06-error-handling.md#3-inference-diagnostics-swallow-some-invalid-request-construction-errors),
+[Testing 4](08-testing.md#4-error-tests-miss-composed-causes-and-pre-dispatch-diagnostic-failures).
+
+## 10. Remove Quadratic Requirement-Evidence Union Work
+
+**Confirmed algorithmic inefficiency.** A left-heavy recognized condition can
+repeatedly copy growing missing-key evidence sets, producing quadratic work.
+Use operation-local accumulation or ownership-aware merging while preserving
+classification and conservative inspection coverage. Do not address this by
+adding expression limits or expanding the recognized template language. Entry
+23's traversal cleanup is related but is not a substitute for fixing this
+specific accumulation algorithm.
+
+Sources: [Efficiency 1](02-efficiency.md#1-condition-evidence-unions-can-require-quadratic-work).
+
+## 11. Exercise RunRequest's Privacy Formatting Boundary
+
+**Missing protection, not a demonstrated leak.** Request formatting methods
+have no execution coverage despite owning a privacy-sensitive public boundary.
+Add focused tests with distinct synthetic private markers for the relevant
+value/pointer formatting paths and JSON omission behavior. Assert privacy and
+the documented boundary, not an incidental complete diagnostic or JSON shape.
+
+Sources: [Testing 1](08-testing.md#1-runrequests-privacy-formatting-contract-is-not-exercised).
+
+## 12. Correct Profile Inspection Credential Documentation
+
+**Public-contract documentation defect.** Inspection GoDoc incorrectly makes
+required credentials and an environment variable mutually exclusive. Actual
+behavior permits an explicit profile environment variable while suppressing
+only the backend-derived default when required credentials are selected.
+Correct the field/method GoDoc and consumer example, keeping the established
+configuration behavior and DTO shape.
+
+Sources: [Interfaces 2](09-interfaces.md#2-profile-inspection-misstates-required-credential-semantics).
+
+## 13. Correct The Configured Provider-Map Wire Documentation
+
+**Integration-contract documentation defect.** The wire document says backend
+and profile provider maps replace one another, but configuration recursively
+overlays them; runtime replacement is a separate rule. Align the integration
+summary with the canonical format reference rather than maintaining a second
+conflicting precedence definition. Entry 19 concerns the efficiency of the
+existing overlay, not a change to its semantics.
+
+Sources: [Interfaces 3](09-interfaces.md#3-the-wire-contract-incorrectly-says-configured-provider-maps-replace).
+
+## 14. Remove Stale Identity And Version GoDoc
+
+**Public-contract documentation defect.** Inspection comments still describe
+trimmed profile IDs, and listing comments describe a removed prompt-version
+field. Update them to the implemented literal identity and current collection
+version model. Do not normalize meaningful whitespace or reintroduce a field
+to match obsolete comments.
+
+Sources: [Interfaces 4](09-interfaces.md#4-identity-godoc-still-describes-removed-normalization-and-version-fields).
+
+## 15. Cover The Public Single-Schema-File Option
+
+**Public wiring coverage gap.** Direct filesystem-validator tests do not
+exercise `WithSchemaFile` through engine construction. Add offline temporary
+file cases through public preparation/validation, including option precedence
+and relevant file errors. This is a supported feature, not dead code; entry
+32 removes only an unused private helper result.
+
+Sources: [Testing 2](08-testing.md#2-the-public-single-schema-file-option-has-no-execution-coverage).
+
+## 16. Detach Inert Prepared Handles From The Execution Owner
+
+**Avoidable retention.** Used or discarded prepared handles retain their
+runner, and therefore the engine's captured configuration and collaborators.
+Use a narrower ownership identity or clear the unnecessary owner reference
+safely. Preserve foreign-owner rejection, one-time claim, namespace-view
+behavior, and available details. This is unnecessary retention when a handle
+outlives its engine, not evidence of an ever-growing global leak.
+
+Sources: [Resources 3](05-resources.md#3-inert-prepared-handles-retain-their-entire-execution-owner).
+
+## 17. Bound Repair Feedback's Temporary Serialization
+
+**Avoidable peak allocation.** Feedback formatting serializes potentially
+large diagnostic text before truncating the final message, so its existing
+output bound does not bound scratch memory. Produce bounded encoded feedback
+incrementally, preserving the existing output semantics and complete stored
+validation diagnostics and previous candidate. A new raw-diagnostic input
+limit is not implied.
+
+Sources: [Resources 4](05-resources.md#4-repair-feedbacks-output-bound-does-not-bound-its-temporary-serialization).
+
+## 18. Detach Normalized Session IDs From Large Input Strings
+
+**Avoidable retention.** A short trimmed session ID can retain the allocation
+of a much larger padded caller string. Clone the accepted normalized value at
+its ownership boundary. Preserve trimming, accepted content, code-point
+validation, and resulting wire/hash content rather than adding a raw-input
+size restriction.
+
+Sources: [Resources 5](05-resources.md#5-short-normalized-session-ids-can-pin-large-input-strings).
+
+## 19. Avoid Repeated Descendant Cloning During Provider Overlay
+
+**Construction-time inefficiency.** Recursive overlays repeatedly clone and
+sort overlapping descendant maps, giving work proportional to tree size times
+depth. Existing depth bounds mean this is not unbounded quadratic growth.
+Clone ownership once or use a single traversal while retaining recursive-map,
+null, and array semantics. Coordinate with entry 24 without assuming all JSON
+walkers have interchangeable policies.
+
+Sources: [Efficiency 2](02-efficiency.md#2-recursive-provider-overlay-repeatedly-clones-descendant-subtrees).
+
+## 20. Index Filtered Listings By Namespace
+
+**Repeated inspection overhead.** Filtered listing currently copies and scans
+the complete inventory even when few or no entries match. Capture namespace
+buckets or equivalent ranges so queries operate on the relevant subset.
+Preserve ordering, exact identity matching, cancellation, and caller-owned
+results. A pagination API or runtime cache is not needed for this improvement.
+
+Sources: [Efficiency 3](02-efficiency.md#3-filtered-listings-copy-and-scan-the-complete-inventory).
+
+## 21. Share Schema Compilation And Declare The Runner's Real Dependency
+
+**Maintainability and interface clarity.** Four host-file/`fs.FS`,
+direct/prepared paths repeat the schema compilation lifecycle. The runner also
+declares a validator interface while requiring preparation through a runtime
+assertion. Share the lifecycle behind source-specific adapters and make that
+dependency explicit. Preserve malformed-output short-circuiting before schema
+I/O, source containment, exact numbers, cancellation, and frozen plans. Retain
+independent cross-source tests; do not force whole entry points to be identical.
+
+Sources: [Complexity 1](03-complexity.md#1-consolidate-schema-compilation-and-express-the-runners-actual-interface),
+[Equivalence 1](10-equivalence.md#1-share-the-schema-compilation-lifecycle-across-source-and-execution-paths).
+
+## 22. Decompose Configuration Validation And Materialization
+
+**Maintainability.** Large configuration operations mix record-specific
+validation, resource capture, and effective-profile construction. Extract
+focused helpers with visible inputs and ownership, rather than another generic
+validation framework. Preserve final-handoff validation, presence/null rules,
+origins, and meaningful fail-fast/error-precedence behavior. Do not turn
+incidental ordering of independent validation failures into a new guarantee. The configuration-boundary ADR remains
+authoritative; this is not a proposal to inspect raw ConfigKit sources.
+
+Sources: [Complexity 2](03-complexity.md#2-decompose-configuration-validation-and-resource-materialization).
+
+## 23. Clarify Requirement Traversal And Share Branch Orchestration
+
+**Maintainability.** Separate recognition/evaluation from reachability and
+evidence bookkeeping, particularly the repeated `if`/`with` branch protocol.
+Keep their scope differences explicit and retain independent comparisons with
+the actual Go renderer. Preserve conservative partial coverage and inactive
+computed-name behavior. This does not justify a general template interpreter
+or stronger execution-readiness promises from requirements inspection.
+
+Sources: [Complexity 3](03-complexity.md#3-untangle-requirement-traversal-expression-evaluation-and-evidence),
+[Equivalence 3](10-equivalence.md#3-share-ifwith-requirement-branch-orchestration-not-template-execution).
+
+## 24. Centralize Compatible JSON Copying Without Conflating Policies
+
+**Maintainability and ownership clarity.** The facade maintains a separate
+reflection traversal for already-normalized values despite an existing
+`internal/jsonvalue` owner. Consolidate genuinely compatible structural
+copying while retaining DTO projection at the facade and distinct source
+encoding/runtime validation policies. Preserve concrete types, nils, exact
+numbers, schema keys, and ownership. Infallible projection must not acquire
+new errors, panics, or silent zero-value fallbacks. Entries 3 and 5 remain
+separate source-conversion defects, not reasons to reject pre-handoff sources.
+
+Sources: [Complexity 4](03-complexity.md#4-clarify-ownership-and-policies-across-the-json-value-walkers),
+[Equivalence 2](10-equivalence.md#2-centralize-structural-copying-of-already-normalized-json-values).
+
+## 25. Separate Attempt Accounting From Generation Control Flow
+
+**Maintainability.** Generation coordination interleaves dispatch/retry policy
+with parallel return values and repeated evidence/observation projection.
+Introduce a small attempt result and focused accounting helpers. Preserve
+built-in versus injected-client differences, permit release before callbacks
+or backoff, frozen requests, existing retry classification, private evidence,
+and known/unknown usage distinctions. Do not collapse legitimately different
+dispatch strategies merely to reduce lines.
+
+Sources: [Complexity 5](03-complexity.md#5-extract-attempt-accounting-from-generation-control-flow).
+
+## 26. Move OpenRouter Authentication Protocol Details Into Its Adapter
+
+**Architectural ownership.** Use-case code currently interprets OpenRouter's
+account-response fields and provider-relative endpoint while the outbound
+adapter owns HTTP execution. Place protocol construction/recognition with the
+outbound integration; keep diagnostic selection, applicability, cancellation,
+completion, and neutral reporting in the use case. Preserve credential/body
+privacy and anonymous probing. No plugin framework, reverse dependency, or
+blanket prohibition on HTTP facts is needed. Coordinate the body-bound fix
+in entry 2, which is a distinct behavioral change.
+
+Sources: [Boundaries 2](11-boundaries.md#2-put-openrouter-authentication-protocol-details-behind-the-outbound-adapter).
+
+## 27. Narrow Smoke-Value Copy Helpers To Real Consumers
+
+**Ownership cleanup.** A request cloner has only its self-test as a caller;
+the full-report cloner is used to detach only a target. Express that actual
+copy operation directly and retire unused broader cloning after checking
+callers. Keep facade request/report projection, active backend cloning,
+provider-map errors, ownership tests, and redaction. Do not introduce shared
+public/internal DTOs or an additional full-report copy.
+
+Sources: [Boundaries 3](11-boundaries.md#3-narrow-smoke-value-copy-helpers-to-their-actual-ownership-boundaries).
+
+## 28. Give Bundled Catalog Expectations One Test Owner
+
+**Test duplication.** Complete bundled profile/model/provider expectations are
+repeated across several test layers. Keep one independently specified asset
+oracle, with smaller tests for distinct loading, wire, and diagnostic behavior.
+Use controlled fixtures where bundled churn is irrelevant and structural
+assertions rather than formatted-map strings. Retain meaningful bundled
+coverage instead of deriving every expected value from the loader under test.
+
+Sources: [Testing 7](08-testing.md#7-bundled-catalog-expectations-have-multiple-test-owners).
+
+## 29. Stop Pinning The Opaque Artifact Hash Algorithm In Tests
+
+**Over-specified tests.** Artifact tests and a benchmark assert a literal hash
+algorithm despite the public value being opaque. Assert the documented
+equality/ownership relationships and actual byte/size behavior instead; keep
+benchmark setup outside its timed work. Coordinate with entry 31, but do not
+turn today's hash representation into a new compatibility guarantee.
+
+Sources: [Testing 6](08-testing.md#6-artifact-tests-and-benchmarks-pin-an-explicitly-opaque-hash-algorithm).
+
+## 30. Remove Unconsumed Backend Presence Bookkeeping
+
+**Low-priority state cleanup.** Backend field-state maps and
+`QueueCapacitySet` have no runtime reader; effective values and ConfigKit
+already own queue defaulting and provenance. Remove only these unused records
+and adjust tests to assert effective behavior or operator inspection. Keep
+profile/input presence tracking, explicit zero capacity, and derived defaults.
+
+Sources: [Dead code 1](12-deadcode.md#1-remove-backend-presence-bookkeeping-with-no-runtime-consumer).
+
+## 31. Retire Duplicate Fields In Captured Prompt Definitions
+
+**Low-priority representation cleanup.** Captured definitions duplicate the
+output format and carry an unused runtime message-file field after content
+materialization. Use the captured output contract and text as their respective
+sources of truth. Preserve real YAML `content_file` loading, origins, and the
+request-override default rule. Removing serialized internal fields changes
+current definition hashes: retain documented opaque equality relationships,
+not an invented bit-identical encoding promise.
+
+Sources: [Dead code 2](12-deadcode.md#2-retire-legacy-fields-in-the-captured-prompt-representation).
+
+## 32. Remove Unused Private Diagnostic State And Helper Results
+
+**Low-priority cleanup.** A private reachability outcome echoes an endpoint
+that report assembly never reads; the schema-file helper returns a directory
+that its only caller discards. Remove the unused state/result while preserving
+actual sanitized request/report evidence, directory calculation, file checks,
+and public single-file schema support. Keep privacy tests against the values
+that real execution consumes.
+
+Sources: [Dead code 3](12-deadcode.md#3-remove-unused-diagnostic-state-and-a-discarded-helper-result).
+
+## 33. Remove Orphan Internal Helpers And Their Self-Tests
+
+**Low-priority cleanup.** `filecatalog.RelativePath` and
+`resourceNamespaceFromID` have only their own tests as callers, and the runner
+test's `stringPtr` helper is entirely unused. Remove those narrow surfaces
+without replacement, preserving real discovery, identity qualification, and
+namespace inventory/filtering coverage. Reachability from examples alone is
+not a reason to remove optional public APIs.
+
+Sources: [Dead code 4](12-deadcode.md#4-remove-orphan-helpers-whose-only-callers-are-their-own-tests).
+
+## 34. Decide Whether To Retain The Legacy Capacity Client Wrapper
+
+**Decision-dependent architectural cleanup.** The old capacity client wrapper
+is used by tests, whereas production dispatch uses the generation coordinator.
+It is also explicitly retained in current internal documentation; the dead-code
+audit does not classify it as obsolete functionality. If that retention decision
+is revised, exercise the production coordinator in runner tests and move any
+needed one-shot capacity scaffolding into tests before retiring the wrapper.
+Preserve FIFO, cancellation, independent-pool, and panic-release coverage.
+This is not a second scheduling algorithm or a confirmed concurrency defect.
+
+Sources: [Boundaries 1](11-boundaries.md#1-retire-the-test-only-capacity-client-as-a-production-scheduling-surface),
+[Dead-code retention qualification](12-deadcode.md#intentional-retention-and-related-findings).
+
+## Open Decisions
+
+Work through the decisions below in this order. These are unresolved choices,
+not additional findings or approved changes to public contracts. A decision
+blocks its dependent remediation, not unrelated fixes. In particular, updating
+the maintainer toolchain need not wait for the consumer-minimum decision.
+Implementation details and optional cleanup do not become release gates merely
+by appearing in this catalog.
+
+### Decision 1: ArtifactReader Result Ownership And Lifetime
+
+**Related finding:** [4](#4-resolve-the-artifactreader-buffer-ownership-contract).
+
+**Question:** What ownership does a successful `Read` transfer, and when may
+an injected reader mutate or reuse the returned storage?
+
+**Recommendation:** Transfer an independent result to PromptKit on successful
+return. The reader must stop mutating the returned artifact and its body,
+including through retained aliases. Fresh per-invocation storage is the simplest
+implementation; immutable shared data can remain shared if no collaborator
+mutates it. This removes the impossible immediate-reuse permission without
+adding synchronization or a release protocol to the interface. Explain the
+consumer impact as a deliberate pre-1.0 contract correction.
+
+**Alternative:** Permit borrowing until the enclosing preparation/execution
+operation returns. This can support application-managed reuse, but a reader
+cannot generally observe that lifetime and concurrent calls need independent
+storage or caller coordination. If reuse immediately after `Read` returns is
+required, an explicit acknowledgement/release API is another viable design;
+it enables pooling at the cost of a larger interface and lifecycle change.
+
+**Resolve before implementation:** Select the ownership/lifetime rule and update
+GoDoc, adapter assumptions, and the regression scenario together. Another
+unsynchronized copy is not a remedy for the current permission.
+
+### Decision 2: Provider Submission Of Multi-Resource Schemas
+
+**Related findings:** [6](#6-make-provider-native-schema-constraints-self-contained)
+and [21](#21-share-schema-compilation-and-declare-the-runners-real-dependency).
+
+**Question:** How should native structured output represent a schema whose
+local validation depends on external resources?
+
+**Recommendation:** First investigate packaging the operation's captured
+resource graph into a self-contained provider constraint. Check identifiers,
+relative references, anchors, recursion, and supported dialects before choosing
+an encoding. Prefer correct packaging when it fits a proportionate scope,
+while keeping resource completeness distinct from any provider's supported
+schema subset. A standalone offline validation of the actual emitted constraint
+should establish the selected packaging behavior.
+
+**Alternative:** Offer an explicit local-only validation choice when native
+submission cannot represent the resource graph. This preserves useful local
+validation and avoids building a general schema transformation system, but
+forgoes native generation constraints for that selected mode and needs a clear
+consumer-facing selection mechanism. Silently dropping native constraints is
+not equivalent to an explicit choice.
+
+**Resolve before implementation:** Establish packaging feasibility and select
+the submission/fallback policy, including behavior for unsupported cases.
+Coordinate the compilation refactor with that result; it does not itself fix
+provider packaging or establish universal provider acceptance.
+
+### Decision 3: Authentication Response Budget And Overflow Outcome
+
+**Related findings:** [2](#2-bound-authentication-probe-response-consumption)
+and [26](#26-move-openrouter-authentication-protocol-details-into-its-adapter).
+
+**Question:** What finite response budget should the authentication adapter
+apply, and how should a probe report exceeding it?
+
+**Recommendation:** Select a conservative fixed budget for this small protocol,
+check actual streamed bytes regardless of declared length, and close on overflow
+without unbounded draining. Report the diagnostic failure with acceptance
+unknown, retaining any independently observed HTTP facts. Agree the exact byte
+count and its representation in the existing diagnostic report before coding;
+the catalog does not select either implicitly.
+
+**Alternative:** Expose a caller-configurable budget with a finite default.
+This accommodates unusual legitimate responses but expands configuration and
+validation surface. It is worthwhile if a concrete consumer requirement cannot
+be served by an appropriately chosen fixed budget; it is not necessary merely
+to address the demonstrated unbounded read.
+
+**Resolve before implementation:** Choose the size, overflow classification,
+and whether configurability is warranted. Keep this fix independent of adapter
+relocation so architectural cleanup does not delay the resource remedy.
+
+### Decision 4: Exact Numbers And The Source-Conversion Boundary
+
+**Related findings:** [3](#3-avoid-exponential-expansion-during-convenience-source-conversion),
+[5](#5-preserve-exact-numbers-in-programmatic-provider-sources), and
+[24](#24-centralize-compatible-json-copying-without-conflating-policies).
+
+**Question:** Which exact numeric values can the convenience helpers preserve
+through the current ConfigKit mapping boundary, and what happens to others?
+
+**Recommendation:** Preserve supported values through the complete conversion,
+load, and outbound-encoding path; defer an unsupported-value error when exact
+preservation is unavailable. Compare numeric meaning rather than spelling or
+successful parsing alone. Address shared-graph expansion in the same adapter
+with completed-conversion memoization, distinct cycle tracking, and correct
+slice-view identity, preserving source capture and deferred failures.
+
+**Alternative:** Request upstream support for additional exact numeric/value
+forms, then simplify or remove the relevant PromptKit conversion. This may
+provide a cleaner long-term boundary and broader preservation, but introduces
+an upstream dependency and should not postpone fixing silent rounding or
+exponential expansion. Runtime JSON copying remains a separate policy.
+
+**Resolve before implementation:** Establish the supported numeric round-trip
+cases and whether broader upstream support is needed now. Do not substitute
+numeric-looking strings or introduce a second raw-source validation system.
+
+### Decision 5: Consumer Go Minimum After The Toolchain Update
+
+**Related finding:** [1](#1-update-the-audited-maintainer-toolchain-for-applicable-security-fixes).
+
+**Question:** Should the security fixes also change PromptKit's declared minimum
+consumer Go version?
+
+**Recommendation:** Update the maintainer toolchain and repeat the focused
+assessment promptly; initially keep the consumer minimum decision separate.
+Advise affected consumers to rebuild with a patched toolchain. This recognizes
+that PromptKit publishes source and that downstream build/transport choices
+control the deployed exposure.
+
+**Alternative:** Raise the declared minimum to the selected patched release.
+This makes the supported floor clearer and steers module consumers toward the
+fixed toolchain, but changes compatibility and does not repair already-built
+applications. Assess it under the existing release policy rather than treating
+it as an automatic consequence of the scanner output.
+
+**Resolve before changing the minimum:** Confirm current advisory applicability,
+select the patched version, and decide whether changing the supported consumer
+floor is justified. No new recurring release gate is implied.
+
+### Decision 6: Retention Of The Legacy Capacity Client Wrapper
+
+**Related finding:** [34](#34-decide-whether-to-retain-the-legacy-capacity-client-wrapper).
+
+**Question:** Does the wrapper still justify a production surface for its
+currently documented test-focused purpose?
+
+**Recommendation:** Retire the production wrapper after moving the runner's
+integration test to the production generation coordinator. Keep direct manager
+coverage and any useful one-shot scaffolding in tests. This reduces maintenance
+and makes assembled scheduling tests reflect production while preserving the
+actual FIFO, cancellation, independent-pool, and panic-release scenarios.
+
+**Alternative:** Retain it if a concrete internal use justifies the small
+adapter. This avoids test migration and preserves that convenient composition,
+but keeps a second assembly surface to understand and maintain. Existing
+self-tests alone do not establish a future production need.
+
+**Resolve before removal:** Confirm the intended retention decision and test
+ownership. This cleanup can be deferred without blocking the behavioral fixes;
+it is not evidence of a second scheduler or a concurrency defect.

docs/roadmap/deferred.md

@@ -42,20 +42,21 @@ checkpoint equality or when a broader semantic identity design is selected.
 ### Eager Source Validation
 
 **Reason for deferral:** Exact prompt and profile inspection may already
-provide a sufficiently small validation surface. Experience from downstream
-adoption should establish whether an engine-wide operation would add enough
-value to justify its broader contract.
+provide a sufficiently small validation surface alongside construction-time
+definition validation. Experience from downstream adoption should establish
+whether a separate operation would add enough value to justify its broader
+contract.
 
-Promptkit could provide an explicit offline operation that discovers and
-structurally validates configured prompt, profile, and schema sources without
-model generation. The normal `NewEngine` path would remain lazy.
+Promptkit could provide an explicit offline operation that also checks schema
+resources without model generation. `NewEngine` already validates effective
+definitions and captures prompt content; schema resources remain
+request-dependent.
 
-An eager operation would need coherent handling for duplicate prompt IDs and
-versions, strict YAML decoding, referenced content files, profile/backend
-membership, schema syntax and transitive references, context cancellation,
-and source-specific public errors. Credential declarations must remain
-separate from credential values; checking current environment availability,
-if supported at all, should be an explicit option and must not expose secrets.
+Such an operation would need coherent handling for schema syntax and transitive
+references, context cancellation, and source-specific public errors. Credential
+declarations must remain separate from credential values; checking current
+environment availability, if supported at all, should be an explicit option
+and must not expose secrets.
 
 Reconsider this idea after downstream use of `InspectPrompt`,
 `InspectProfile`, and fixture-based preparation demonstrates a concrete gap.

docs/roadmap/future.md

@@ -38,28 +38,8 @@ consumers.
 
 ## Ideas
 
-### Public bounded output repair
-
-Promptkit should make its bounded output-repair capability available through
-the public engine. A consumer should be able to request a limited number of
-corrective generation attempts when JSON or JSON Schema output fails content
-validation, without having to reproduce Promptkit's generation, validation,
-capacity, and result-accounting orchestration.
-
-- Repair is validation recovery, not a general provider retry, failover, or
-  backoff policy. Transport failures, cancellation, and operational schema or
-  validation errors must retain their ordinary error behavior.
-- Repair must stop after the first valid result or the configured attempt
-  bound. Exhausting the bound should preserve the final invalid result and its
-  validation diagnostics rather than inventing success.
-- Initial generation and every repair attempt must use the same resolved
-  backend, effective execution settings and presence semantics, session,
-  credential boundary, structured-output contract, and backend-capacity
-  policy.
-- Results should report the number of repair attempts and cumulative usage for
-  every model call made by the run.
-- Ordinary and prepared execution should expose coherent behavior, including
-  cancellation, frozen prepared state, error identity, and capacity lifetime.
+No unselected ideas are currently recorded here. Deliberately postponed ideas
+remain in the [deferred catalog](deferred.md).
 
 ## Entry Format
 

engine.go

@@ -6,27 +6,17 @@ import (
 	"fmt"
 	"io/fs"
 	"net/http"
-	"os"
-	"path/filepath"
-	"strings"
 	"time"
 
-	artifactadapter "gitea.maximumdirect.net/eric/promptkit/internal/artifact"
-	"gitea.maximumdirect.net/eric/promptkit/internal/backend"
-	"gitea.maximumdirect.net/eric/promptkit/internal/capacity"
-	"gitea.maximumdirect.net/eric/promptkit/internal/defaults"
-	"gitea.maximumdirect.net/eric/promptkit/internal/domain"
-	"gitea.maximumdirect.net/eric/promptkit/internal/llm"
-	"gitea.maximumdirect.net/eric/promptkit/internal/profile"
-	"gitea.maximumdirect.net/eric/promptkit/internal/profile/builtin"
-	"gitea.maximumdirect.net/eric/promptkit/internal/prompt"
-	"gitea.maximumdirect.net/eric/promptkit/internal/promptdef"
-	"gitea.maximumdirect.net/eric/promptkit/internal/usecase"
-	"gitea.maximumdirect.net/eric/promptkit/internal/validate"
+	"go.maximumdirect.net/configkit"
+	"go.maximumdirect.net/promptkit/internal/domain"
+	internalengine "go.maximumdirect.net/promptkit/internal/engine"
+	"go.maximumdirect.net/promptkit/internal/observation"
+	"go.maximumdirect.net/promptkit/internal/usecase"
 )
 
 // ErrInvalidConfig identifies invalid engine construction, including missing
-// required configuration, invalid options or backend registrations, and a nil
+// effective configuration, invalid options, and a nil
 // Engine receiver.
 var ErrInvalidConfig = errors.New("invalid engine configuration")
 
@@ -34,9 +24,8 @@ var (
 	// ErrInvalidRequest identifies a request whose required values, overrides,
 	// credentials, or effective settings are invalid.
 	ErrInvalidRequest = errors.New("invalid run request")
-	// ErrPromptNotFound identifies a requested prompt ID or version that is not
-	// present in the selected prompt source. It does not also match
-	// ErrPromptLoad.
+	// ErrPromptNotFound identifies a requested prompt ID that is not present in
+	// the captured configuration. It does not also match ErrPromptLoad.
 	ErrPromptNotFound = errors.New("prompt not found")
 	// ErrProfileNotFound identifies a selected profile ID that is absent from
 	// every configured profile source. It does not also match ErrProfileLoad.
@@ -45,13 +34,13 @@ var (
 	// nor the selected prompt's default profile is present. Such an error also
 	// matches ErrInvalidRequest.
 	ErrProfileRequired = errors.New("profile selection is required")
-	// ErrPromptLoad identifies a failure to read, decode, validate, select, or
-	// hash a prompt definition, except for the not-found case represented by
-	// ErrPromptNotFound.
+	// ErrPromptLoad identifies a prompt selection failure other than the
+	// not-found case represented by ErrPromptNotFound. Invalid definitions
+	// instead fail construction with ErrInvalidConfig.
 	ErrPromptLoad = errors.New("failed to load prompt definition")
-	// ErrProfileLoad identifies a failure to read, decode, validate, or select
-	// an execution profile or resolve its backend, except for the profile
-	// not-found case represented by ErrProfileNotFound.
+	// ErrProfileLoad identifies a profile selection or resolution failure other
+	// than the not-found case represented by ErrProfileNotFound. Invalid
+	// definitions instead fail construction with ErrInvalidConfig.
 	ErrProfileLoad = errors.New("failed to load execution profile")
 	// ErrAPIKeyEnvMissing identifies an explicitly required APIKeyEnv whose
 	// environment variable is unset or empty after direct RunRequest.APIKey
@@ -73,37 +62,60 @@ var (
 	// [GenerationError]. Errors returned by an injected LLMClient remain
 	// available through errors.Is.
 	ErrLLMGenerate = errors.New("failed to generate output")
+	// ErrTransport identifies a failed HTTP request by the built-in model
+	// client. Generation errors in this category also match ErrLLMGenerate and
+	// retain the transport cause for errors.Is and errors.As. It does not imply
+	// that the provider received no work.
+	ErrTransport = errors.New("model transport failed")
+	// ErrMalformedResponse identifies an invalid successful model response or
+	// a nil successful response from an injected client. Generation errors in
+	// this category also match ErrLLMGenerate.
+	ErrMalformedResponse = errors.New("malformed model response")
+	// ErrRetriesExhausted identifies a built-in generation sequence whose
+	// positive transient retry budget ended with another eligible HTTP status.
+	// RetryExhaustedError retains the final cause and actual attempt count.
+	ErrRetriesExhausted = errors.New("transient model retries exhausted")
 	// ErrValidation identifies an operational failure to load or compile a
 	// schema or validate output. A completed validation whose Status is
 	// ValidationFailed is returned in RunResult without this error.
 	ErrValidation = errors.New("failed to validate output")
 )
 
-// Engine inspects prompts and profiles and prepares and runs Promptkit prompt
-// requests.
+// Engine inspects prompts and profiles, enumerates prompt requirements, and
+// prepares and runs Promptkit prompt requests.
 //
-// An Engine is safe for concurrent calls to [Engine.InspectPrompt],
-// [Engine.InspectProfile], [Engine.Prepare], [Engine.PrepareExecution],
-// [Engine.Run], and [Engine.RunPrepared]. Each Engine owns independent
-// backend-capacity pools that coordinate Run and RunPrepared admission and
-// model generation. Injected collaborators may still be invoked concurrently
-// across different backend pools or for unlimited backends.
+// An Engine is safe for concurrent calls to [Engine.WithNamespaces],
+// [Engine.InspectConfiguration],
+// [Engine.InspectPrompt],
+// [Engine.InspectRequirements], [Engine.InspectProfile], [Engine.Prepare], [Engine.PrepareExecution],
+// [Engine.Run], [Engine.RunPrepared], [Engine.SmokeTest], [Engine.ListProfiles],
+// [Engine.ListPrompts], [Engine.ListBackends], [Engine.ListNamespaces], and
+// [Engine.ListCollections]. Engines constructed separately by [NewEngine] own
+// independent backend-capacity pools that coordinate Run, RunPrepared, and
+// inference SmokeTest admission and model generation. Namespace views returned
+// by [Engine.WithNamespaces] share their underlying engine's pools, clients,
+// and prepared-execution owner. Injected collaborators may still be invoked
+// concurrently across different backend pools or for unlimited backends.
 type Engine struct {
-	runner *usecase.Runner
+	internal *internalengine.Engine
 }
 
-// Config selects the directory-backed sources and built-in model-client
-// transport used by [NewEngine]. Config has no stable JSON representation.
+// ConfigurationSources selects ordered ConfigKit underlays and an optional
+// highest-priority main input. By default a bundled OpenRouter underlay precedes
+// the caller underlays; [WithBundledOpenRouter] can disable it. Inputs within
+// each underlay strictly compose; later underlays overlay earlier ones. A nil Main uses
+// ConfigKit's empty source. A non-nil Main is passed through unchanged and
+// may fail construction if invalid.
+type ConfigurationSources struct {
+	Underlays []configkit.Underlay
+	Main      *configkit.Input
+}
+
+// Config selects configuration sources and built-in model-client transport
+// used by [NewEngine]. Config has no stable JSON representation.
 type Config struct {
-	// PromptDir is the directory searched recursively for prompt definitions.
-	// It is required unless a WithPromptFS or WithPromptFile option supplies the
-	// prompt source.
-	PromptDir string
-	// ProfileDir is an optional ordinary configured source whose profiles take
-	// precedence over application fallback and embedded built-in profiles. An
-	// empty value selects the lower-precedence sources unless a profile-source
-	// option supplies the ordinary source.
-	ProfileDir string
+	// Sources combines maintained definitions with caller-selected inputs.
+	Sources ConfigurationSources
 	// SchemaDir is the root for JSON Schema files. An empty value uses the
 	// current directory. WithSchemaFS or WithSchemaFile replaces this source.
 	SchemaDir string
@@ -116,17 +128,19 @@ type Config struct {
 	// Timeout or the 10-minute default. The supplied client is not mutated. This
 	// field is ignored when WithLLMClient is used.
 	HTTPClient *http.Client
+	// TransientRetries selects additional built-in HTTP attempts after eligible
+	// provider statuses. Nil defaults to two; zero disables retries. Negative
+	// values are invalid. Injected LLMClient implementations are never retried
+	// automatically.
+	TransientRetries *int
 }
 
 // Option customizes engine construction.
 //
 // NewEngine applies options in argument order and ignores nil options. Within
-// each prompt-source, ordinary-profile-source, fallback-profile-source,
-// in-memory-profile, schema-source, model-client, and artifact-reader
-// category, the last non-nil valid option replaces earlier options in that
-// category. WithBackend is the additive exception: unique registrations
-// accumulate, and a repeated backend ID is an error rather than a replacement.
-// An invalid option fails construction even if a later option would replace it.
+// each schema-source, model-client, and artifact-reader category, the last
+// non-nil valid option replaces earlier options. An invalid option fails
+// construction even if a later option would replace it.
 type Option interface {
 	apply(*engineOptions) error
 }
@@ -138,20 +152,16 @@ func (f optionFunc) apply(options *engineOptions) error {
 }
 
 type engineOptions struct {
-	llmClient             llm.Client
-	artifactReader        artifactadapter.Reader
-	promptDefs            promptdef.Repository
-	profiles              profile.Repository
-	fallbackProfiles      profile.Repository
-	memoryProfiles        profile.Repository
-	backends              []domain.Backend
-	validator             validate.Validator
-	promptSource          bool
-	profileSource         bool
-	fallbackProfileSource bool
-	memorySource          bool
-	validatorSource       bool
-	artifactSource        bool
+	internal internalengine.Options
+}
+
+// WithBundledOpenRouter enables or disables the built-in OpenRouter underlay.
+// It is enabled by default; when repeated, the last option wins.
+func WithBundledOpenRouter(enabled bool) Option {
+	return optionFunc(func(options *engineOptions) error {
+		options.internal.SetBundledOpenRouter(enabled)
+		return nil
+	})
 }
 
 // WithLLMClient replaces the built-in model client used by [Engine.Run] and
@@ -167,7 +177,7 @@ func WithLLMClient(client LLMClient) Option {
 		if client == nil {
 			return ErrInvalidConfig
 		}
-		options.llmClient = publicLLMClientAdapter{client: client}
+		options.internal.SetLLMClient(publicLLMClientAdapter{client: client})
 		return nil
 	})
 }
@@ -182,142 +192,7 @@ func WithArtifactReader(reader ArtifactReader) Option {
 		if reader == nil {
 			return ErrInvalidConfig
 		}
-		options.artifactReader = publicArtifactReaderAdapter{reader: reader}
-		options.artifactSource = true
-		return nil
-	})
-}
-
-// WithPromptFS loads prompt definitions from fsys under root.
-//
-// The source uses the same strict prompt YAML rules as configured prompt
-// directories, and prompt content_file paths resolve within this source.
-// fsys must be non-nil and root must be non-empty; otherwise NewEngine fails
-// with ErrInvalidConfig. This option replaces Config.PromptDir and earlier
-// prompt-source options.
-func WithPromptFS(fsys fs.FS, root string) Option {
-	return optionFunc(func(options *engineOptions) error {
-		if fsys == nil {
-			return ErrInvalidConfig
-		}
-		if strings.TrimSpace(root) == "" {
-			return ErrInvalidConfig
-		}
-		options.promptDefs = promptdef.NewFSRepository(fsys, root)
-		options.promptSource = true
-		return nil
-	})
-}
-
-// WithPromptFile loads prompt definitions from the single prompt file at path.
-//
-// Relative prompt content_file paths resolve from the file's directory. path
-// must name an existing non-directory file when NewEngine applies the option.
-// This option replaces Config.PromptDir and earlier prompt-source options.
-func WithPromptFile(path string) Option {
-	return optionFunc(func(options *engineOptions) error {
-		fsys, root, err := fileSource(path)
-		if err != nil {
-			return err
-		}
-		options.promptDefs = promptdef.NewFileRepository(fsys, root, filepath.Dir(path))
-		options.promptSource = true
-		return nil
-	})
-}
-
-// WithProfileFS loads execution profiles from fsys under root.
-//
-// Profiles from this ordinary configured source take precedence over
-// application fallback and built-in profiles. Profile YAML must use api_key_env
-// for environment-based credentials; raw API keys are rejected. fsys must be
-// non-nil and root must be non-empty; otherwise NewEngine fails with
-// ErrInvalidConfig. This option replaces Config.ProfileDir and earlier file or
-// FS profile-source options, but remains below WithProfiles in precedence.
-func WithProfileFS(fsys fs.FS, root string) Option {
-	return optionFunc(func(options *engineOptions) error {
-		if fsys == nil {
-			return ErrInvalidConfig
-		}
-		if strings.TrimSpace(root) == "" {
-			return ErrInvalidConfig
-		}
-		options.profiles = profile.NewFSRepository(fsys, root)
-		options.profileSource = true
-		return nil
-	})
-}
-
-// WithProfileFile loads execution profiles from the single profile file at path.
-//
-// The profile takes precedence over application fallback and built-in profiles.
-// Profile YAML must use api_key_env for environment-based credentials; raw API
-// keys are rejected. path must name an existing non-directory file when
-// NewEngine applies the option. This option replaces Config.ProfileDir and
-// earlier file or FS profile-source options, but remains below WithProfiles in
-// precedence.
-func WithProfileFile(path string) Option {
-	return optionFunc(func(options *engineOptions) error {
-		fsys, root, err := fileSource(path)
-		if err != nil {
-			return err
-		}
-		options.profiles = profile.NewFSRepository(fsys, root)
-		options.profileSource = true
-		return nil
-	})
-}
-
-// WithFallbackProfileFS supplies application-owned fallback profile
-// definitions from fsys under root.
-//
-// Profile lookup checks, in order, profiles supplied by WithProfiles; the
-// ordinary configured source selected by WithProfileFile, WithProfileFS, or
-// Config.ProfileDir; this fallback source; and Promptkit's embedded built-in
-// profiles. Each source supplies a complete profile definition; profile fields
-// are not merged between sources. Only an absent profile ID proceeds to the
-// next source. A matching read, parse, duplicate, validation, or credential
-// format failure stops resolution.
-//
-// Files use the ordinary strict profile YAML and api_key_env credential rules.
-// Loading and validation are lazy: NewEngine validates this option's arguments
-// but does not read profile files. fsys must be non-nil and root must be
-// nonblank; otherwise NewEngine returns an error matching ErrInvalidConfig.
-// Repeating this option replaces the earlier valid fallback source.
-//
-// This option controls profile-definition lookup, not provider or generation
-// failover.
-func WithFallbackProfileFS(fsys fs.FS, root string) Option {
-	return optionFunc(func(options *engineOptions) error {
-		if fsys == nil {
-			return ErrInvalidConfig
-		}
-		if strings.TrimSpace(root) == "" {
-			return ErrInvalidConfig
-		}
-		options.fallbackProfiles = profile.NewFSRepository(fsys, root)
-		options.fallbackProfileSource = true
-		return nil
-	})
-}
-
-// WithProfiles configures in-memory profiles that take precedence over
-// ordinary configured, application fallback, and built-in profiles.
-//
-// NewEngine locally validates and copies every profile. IDs must be unique
-// within one call. An invalid local definition, duplicate ID, or unsupported
-// ExtraParams value makes construction fail with ErrInvalidConfig. A derived
-// profile's base reference and resolved target completeness are checked when it
-// is selected or inspected. Repeating WithProfiles replaces the complete
-// earlier in-memory set rather than merging it.
-func WithProfiles(profiles ...Profile) Option {
-	return optionFunc(func(options *engineOptions) error {
-		repo, err := newMemoryProfileRepository(profiles)
-		if err != nil {
-			return err
-		}
-		options.memoryProfiles = repo
-		options.memorySource = true
+		options.internal.SetArtifactReader(publicArtifactReaderAdapter{reader: reader})
 		return nil
 	})
 }
@@ -330,15 +205,7 @@ func WithProfiles(profiles ...Profile) Option {
 // replaces Config.SchemaDir and earlier schema-source options.
 func WithSchemaFS(fsys fs.FS, root string) Option {
 	return optionFunc(func(options *engineOptions) error {
-		if fsys == nil {
-			return ErrInvalidConfig
-		}
-		if strings.TrimSpace(root) == "" {
-			return ErrInvalidConfig
-		}
-		options.validator = validate.NewFSValidator(fsys, root)
-		options.validatorSource = true
-		return nil
+		return options.internal.SetSchemaFS(fsys, root)
 	})
 }
 
@@ -349,28 +216,26 @@ func WithSchemaFS(fsys fs.FS, root string) Option {
 // replaces Config.SchemaDir and earlier schema-source options.
 func WithSchemaFile(path string) Option {
 	return optionFunc(func(options *engineOptions) error {
-		fsys, root, err := fileSource(path)
-		if err != nil {
-			return err
-		}
-		options.validator = validate.NewFSValidator(fsys, root)
-		options.validatorSource = true
-		return nil
+		return options.internal.SetSchemaFile(path)
 	})
 }
 
 // NewEngine constructs an Engine from configuration and options.
 //
-// Options are applied in order according to [Option]. PromptDir is required
-// unless a prompt-source option is present. Construction validates option
-// arguments, in-memory profiles, and backend registrations but defers reading
-// and validating prompt, file-backed profile, and schema contents until Prepare
-// or Run needs them.
+// ConfigKit composes caller-selected underlays in their given order, followed
+// by the optional main input, after the default-on bundled OpenRouter underlay.
+// The bundle adds a root backend and profiles to inventories, but no prompts
+// or implicit profile selection. Construction validates the final effective
+// backend, profile, and prompt definitions and captures referenced prompt
+// content. A missing prompt collection is allowed. Schema documents and
+// request artifacts remain runtime resources.
 //
 // NewEngine returns an error matching ErrInvalidConfig for invalid
-// configuration, options, or backend-capacity policies. Each constructed
+// configuration or options, preserving a ConfigKit LoadError when loading
+// fails. Each constructed
 // Engine has independent backend-capacity pools. Construction does not perform
-// model requests or require credentials.
+// model requests or require credentials. A negative Config.TransientRetries
+// is invalid.
 func NewEngine(cfg Config, opts ...Option) (*Engine, error) {
 	var options engineOptions
 	for _, opt := range opts {
@@ -378,118 +243,27 @@ func NewEngine(cfg Config, opts ...Option) (*Engine, error) {
 			continue
 		}
 		if err := opt.apply(&options); err != nil {
-			return nil, fmt.Errorf("%w: %v", ErrInvalidConfig, err)
+			return nil, fmt.Errorf("%w: %w", ErrInvalidConfig, err)
 		}
 	}
 
-	promptDefs := options.promptDefs
-	if !options.promptSource {
-		if strings.TrimSpace(cfg.PromptDir) == "" {
-			return nil, fmt.Errorf("%w: prompt directory is required", ErrInvalidConfig)
-		}
-		promptDefs = promptdef.NewFilesystemRepository(cfg.PromptDir)
-	}
-
-	profiles := newProfileRepository(cfg.ProfileDir, options)
-
-	backendRegistry, err := backend.NewRegistry(options.backends)
+	assembled, err := internalengine.New(
+		internalengine.Config{Sources: internalengine.ConfigurationSources{Underlays: cfg.Sources.Underlays, Main: cfg.Sources.Main}, SchemaDir: cfg.SchemaDir, Timeout: cfg.Timeout, HTTPClient: cfg.HTTPClient, TransientRetries: copyIntPtr(cfg.TransientRetries)},
+		options.internal,
+	)
 	if err != nil {
-		return nil, fmt.Errorf("%w: failed to construct backend registry: %v", ErrInvalidConfig, err)
+		return nil, fmt.Errorf("%w: %w", ErrInvalidConfig, err)
 	}
-
-	capacityManager, err := capacity.NewManager(backendRegistry.CapacityPolicies())
-	if err != nil {
-		return nil, fmt.Errorf("%w: failed to construct backend capacity manager: %v", ErrInvalidConfig, err)
-	}
-
-	validator := options.validator
-	if !options.validatorSource {
-		schemaDir := cfg.SchemaDir
-		if strings.TrimSpace(schemaDir) == "" {
-			schemaDir = defaults.SchemaDirDefault
-		}
-		validator = validate.NewStandardValidator(schemaDir)
-	}
-
-	llmClient := options.llmClient
-	if llmClient == nil {
-		var err error
-		llmClient, err = llm.NewOpenAICompatibleClient(llm.OpenAICompatibleConfig{
-			Timeout:    cfg.Timeout,
-			HTTPClient: cfg.HTTPClient,
-		})
-		if err != nil {
-			return nil, fmt.Errorf("%w: %v", ErrInvalidConfig, err)
-		}
-	}
-	llmClient = capacity.NewClient(capacityManager, llmClient)
-
-	artifacts := options.artifactReader
-	if !options.artifactSource {
-		artifacts = artifactadapter.NewCompositeReader()
-	}
-
-	return &Engine{
-		runner: usecase.NewRunner(
-			promptDefs,
-			profiles,
-			backendRegistry,
-			artifacts,
-			prompt.NewGoRenderer(),
-			llmClient,
-			validator,
-			capacityManager,
-		),
-	}, nil
+	return &Engine{internal: assembled}, nil
 }
 
-func newProfileRepository(profileDir string, options engineOptions) profile.Repository {
-	repository := builtin.NewRepository()
-
-	if options.fallbackProfileSource {
-		repository = profile.NewOverlayRepository(options.fallbackProfiles, repository)
-	}
-
-	if options.profileSource {
-		repository = profile.NewOverlayRepository(options.profiles, repository)
-	} else if strings.TrimSpace(profileDir) != "" {
-		repository = profile.NewOverlayRepository(profile.NewFilesystemRepository(profileDir), repository)
-	}
-
-	if options.memorySource {
-		repository = profile.NewOverlayRepository(options.memoryProfiles, repository)
-	}
-
-	return profile.NewResolvingRepository(repository)
-}
-
-func fileSource(name string) (fs.FS, string, error) {
-	if strings.TrimSpace(name) == "" {
-		return nil, "", ErrInvalidConfig
-	}
-	dir := filepath.Dir(name)
-	base := filepath.Base(name)
-	if base == "." || base == string(filepath.Separator) {
-		return nil, "", ErrInvalidConfig
-	}
-	info, err := os.Stat(name)
-	if err != nil {
-		return nil, "", fmt.Errorf("%w: failed to access source file %q: %v", ErrInvalidConfig, name, err)
-	}
-	if info.IsDir() {
-		return nil, "", fmt.Errorf("%w: source path %q must be a file", ErrInvalidConfig, name)
-	}
-	return os.DirFS(dir), filepath.ToSlash(base), nil
-}
-
-// InspectPrompt resolves one explicit prompt definition without selecting a
-// profile or starting execution work.
+// InspectPrompt resolves one prompt ID without selecting a profile or starting
+// execution work.
 //
-// InspectPrompt requires a nonblank promptID. It passes nonblank promptID and
-// promptVersion values unchanged to the engine's ordinary, case-sensitive
-// prompt selection. An empty version succeeds only when that source has one
-// selected ID; a nonempty version selects one exact ID/version pair. The
-// configured prompt source is used without merging, fallback, or enumeration.
+// InspectPrompt requires a nonblank promptID and performs an exact,
+// case-sensitive lookup in the captured configuration snapshot. A
+// namespace-scoped view qualifies slashless caller IDs before lookup; already
+// qualified IDs pass through.
 //
 // A successful result proves that the selected definition and any referenced
 // message content files were structurally loaded. Inputs are returned in
@@ -497,32 +271,30 @@ func fileSource(name string) (fs.FS, string, error) {
 // resolved. OutputContract is the normalized declared contract, with a JSON
 // Schema path when declared but without loading or compiling that schema.
 // PromptHash is the same opaque equality value as PreparedRun.PromptHash for
-// the selected definition and observed source state; its spelling, length,
+// the selected captured definition; its spelling, length,
 // encoding, algorithm, and security properties are not contracts.
 //
 // This method does not return prompt bodies, templates, source paths, schemas,
 // rendered messages, or execution settings. It does not resolve a profile or
 // credential, read artifacts or schemas, render, validate, admit capacity,
 // contact a provider, or generate model output. The returned PromptInspection
-// and its input slice are caller-owned. Filesystem-backed inspection is a
-// point-in-time lookup and does not freeze a definition for later execution.
+// and its input slice are caller-owned. Inspection does not reload configuration
+// or content files; construct another Engine to observe source changes.
 //
 // A nil Engine returns an error matching ErrInvalidConfig. A blank prompt ID
-// matches ErrInvalidRequest. An absent exact ID or version matches
-// ErrPromptNotFound and not ErrPromptLoad. Malformed, unreadable, duplicate,
-// ambiguous, referenced-content, or hashing failures match ErrPromptLoad.
-// Cancellation during lookup matches ErrPromptLoad while preserving the
-// context error. InspectPrompt returns no partial result on error.
+// matches ErrInvalidRequest. An absent exact ID matches ErrPromptNotFound and
+// not ErrPromptLoad. Invalid definitions and resource failures reject
+// construction instead. Cancellation during lookup preserves the context
+// error. InspectPrompt returns no partial result on error.
 func (e *Engine) InspectPrompt(
 	ctx context.Context,
 	promptID string,
-	promptVersion string,
 ) (*PromptInspection, error) {
-	if e == nil || e.runner == nil {
+	if e == nil || e.internal == nil {
 		return nil, fmt.Errorf("%w: engine is nil", ErrInvalidConfig)
 	}
 
-	inspection, err := e.runner.InspectPrompt(ctx, promptID, promptVersion)
+	inspection, err := e.internal.InspectPrompt(ctx, promptID)
 	if err != nil {
 		return nil, mapPublicError(err)
 	}
@@ -532,10 +304,9 @@ func (e *Engine) InspectPrompt(
 // InspectProfile resolves one explicit profile without selecting a prompt or
 // starting execution work.
 //
-// InspectProfile trims surrounding whitespace from profileID and looks up the
-// resulting nonblank ID exactly and case-sensitively through the engine's
-// in-memory, ordinary configured-source, application fallback, and built-in
-// profile precedence. It applies the framework timeout baseline, selected
+// InspectProfile looks up a nonblank profileID exactly and case-sensitively
+// in the captured effective configuration. It applies the framework timeout
+// baseline, selected
 // backend, and then selected profile to EffectiveModelParams without a request
 // override. BackendID is empty for an endpoint-only profile.
 //
@@ -546,23 +317,22 @@ func (e *Engine) InspectPrompt(
 // absent or blank named environment variable is not an error.
 //
 // The returned ProfileInspection and all nested mutable values are
-// caller-owned. Filesystem-backed inspection is a point-in-time lookup and
-// does not freeze the profile for a later execution. This method does not load
+// caller-owned. Inspection does not reload configuration; construct another
+// Engine to observe source changes. This method does not load
 // a prompt, render, read artifacts or schemas, admit backend capacity, contact
 // a provider, or generate model output.
 //
 // A nil Engine returns an error matching ErrInvalidConfig. A blank profile ID
 // matches ErrInvalidRequest. An absent exact ID matches ErrProfileNotFound and
-// not ErrProfileLoad. Malformed or unreadable profile data, an unknown backend,
-// or an invalid resolved target matches ErrProfileLoad. Cancellation during
-// profile loading matches ErrProfileLoad while preserving the context error.
+// not ErrProfileLoad. Invalid definitions or unknown backends reject
+// construction instead. Cancellation preserves the context error.
 // InspectProfile returns no partial result on error.
 func (e *Engine) InspectProfile(ctx context.Context, profileID string) (*ProfileInspection, error) {
-	if e == nil || e.runner == nil {
+	if e == nil || e.internal == nil {
 		return nil, fmt.Errorf("%w: engine is nil", ErrInvalidConfig)
 	}
 
-	inspection, err := e.runner.InspectProfile(ctx, profileID)
+	inspection, err := e.internal.InspectProfile(ctx, profileID)
 	if err != nil {
 		return nil, mapPublicError(err)
 	}
@@ -570,6 +340,8 @@ func (e *Engine) InspectProfile(ctx context.Context, profileID string) (*Profile
 }
 
 // Prepare resolves and renders a prompt request without calling an LLM.
+// It appends any validated RunRequest.AppendedMessages after rendered
+// definition messages in the returned caller-owned snapshot.
 //
 // Prepare selects the prompt and profile, resolves any selected backend and
 // effective execution settings, resolves the output contract, loads and hashes
@@ -585,9 +357,36 @@ func (e *Engine) InspectProfile(ctx context.Context, profileID string) (*Profile
 // applicable. Cancellation is passed to the active collaborator and is
 // reported in the applicable operation category; no general errors.Is
 // relationship to ctx.Err is promised. Prepare returns no partial result on
-// error.
+// error. It validates RunRequest.TransientRetries but makes no model attempts.
 func (e *Engine) Prepare(ctx context.Context, req RunRequest) (*PreparedRun, error) {
-	if e == nil || e.runner == nil {
+	prepared, err := e.prepareRequest(ctx, req)
+	if err != nil {
+		return nil, err
+	}
+	return fromDomainPreparedRun(prepared), nil
+}
+
+// ValidateRequest reports whether req passes the same preparation path used by
+// [Engine.Prepare] at the time of this call. A nil error means this concrete
+// request passed preparation. It checks only this invocation and returns errors
+// as encountered rather than aggregating them. All supplied inputs follow the
+// same loading behavior as [Engine.Prepare], including inputs unused by the
+// selected template branch. The method does not call an LLM or retain a
+// prepared snapshot or executable handle, but preparation may still perform
+// resource I/O and incur its associated time and memory costs.
+//
+// Its errors and context handling follow [Engine.Prepare], including an error
+// matching ErrInvalidConfig when the Engine is nil. Success does not guarantee
+// provider acceptance or later execution success. Use [Engine.PrepareExecution]
+// when execution must use the snapshot produced by preparation.
+func (e *Engine) ValidateRequest(ctx context.Context, req RunRequest) error {
+	_, err := e.prepareRequest(ctx, req)
+	return err
+}
+
+func (e *Engine) prepareRequest(ctx context.Context, req RunRequest) (*domain.PreparedRun, error) {
+	ctx = observation.WithoutExecutionState(ctx)
+	if e == nil || e.internal == nil {
 		return nil, fmt.Errorf("%w: engine is nil", ErrInvalidConfig)
 	}
 
@@ -596,18 +395,20 @@ func (e *Engine) Prepare(ctx context.Context, req RunRequest) (*PreparedRun, err
 		return nil, fmt.Errorf("%w: %v", ErrInvalidRequest, err)
 	}
 
-	prepared, err := e.runner.Prepare(ctx, domainReq)
+	prepared, err := e.internal.Prepare(ctx, domainReq)
 	if err != nil {
 		return nil, mapPublicError(err)
 	}
-	return fromDomainPreparedRun(prepared), nil
+	return prepared, nil
 }
 
 // PrepareExecution completely prepares a prompt request without calling the
-// configured LLMClient or reserving backend admission capacity.
+// configured LLMClient or reserving backend admission capacity. Validated
+// RunRequest.AppendedMessages are included in the frozen effective messages.
 //
-// The returned opaque handle is bound to this Engine and permits one
-// [Engine.RunPrepared] invocation. Preparation freezes the selected sources,
+// The returned opaque handle is bound to this Engine's underlying execution
+// owner, shared with its namespace views, and permits one [Engine.RunPrepared]
+// invocation in total. Preparation freezes the selected sources,
 // rendered messages, effective settings, inputs, provider structured-output
 // metadata, and validation resources needed by that invocation. The handle
 // retains a direct RunRequest.APIKey only in private execution state;
@@ -617,9 +418,12 @@ func (e *Engine) Prepare(ctx context.Context, req RunRequest) (*PreparedRun, err
 // returns does not invalidate the handle or propagate to RunPrepared.
 // PrepareExecution returns the same error categories as [Engine.Prepare] and
 // returns no handle on error. A nil Engine returns an error matching
-// ErrInvalidConfig.
+// ErrInvalidConfig. The effective transient retry budget is frozen privately
+// for RunPrepared and is absent from Details. A supplied Observer is retained
+// privately for RunPrepared; preparation itself emits no events.
 func (e *Engine) PrepareExecution(ctx context.Context, req RunRequest) (*PreparedExecution, error) {
-	if e == nil || e.runner == nil {
+	ctx = observation.WithoutExecutionState(ctx)
+	if e == nil || e.internal == nil {
 		return nil, fmt.Errorf("%w: engine is nil", ErrInvalidConfig)
 	}
 
@@ -628,7 +432,7 @@ func (e *Engine) PrepareExecution(ctx context.Context, req RunRequest) (*Prepare
 		return nil, fmt.Errorf("%w: %v", ErrInvalidRequest, err)
 	}
 
-	prepared, err := e.runner.PrepareExecution(ctx, domainReq)
+	prepared, err := e.internal.PrepareExecution(ctx, domainReq)
 	if err != nil {
 		return nil, mapPublicError(err)
 	}
@@ -636,28 +440,68 @@ func (e *Engine) PrepareExecution(ctx context.Context, req RunRequest) (*Prepare
 }
 
 // Run prepares a request, invokes the configured LLMClient, and validates the
-// generated output.
+// generated output. Each call uses the Engine's captured configuration and
+// composes a fresh effective prompt with any validated
+// RunRequest.AppendedMessages.
 //
 // A content-validation failure is a successful run whose
-// RunResult.Validation has Status ValidationFailed. An inability to perform
-// validation returns an error matching ErrValidation and no partial result.
-// The public Engine does not perform output repair, so validation is
-// single-pass even when OutputContract.RepairAttempts is positive.
+// RunResult.Validation has Status ValidationFailed. When its output contract
+// has a positive repair budget, a failed eligible validation can make bounded
+// additional model calls and stops at the first valid candidate. Exhaustion
+// returns the final failed validation result with cumulative usage and actual
+// repair attempts. An inability to generate or validate returns an error and
+// no partial result.
+//
+// The built-in client retries eligible provider statuses using the effective
+// Config or request budget. With R additional transient attempts and P repair
+// generations, one run can dispatch at most (R+1)×(P+1) Promptkit-controlled
+// model attempts. Each dispatched attempt receives its own generation and
+// HTTP-client timeout; the caller context governs the full operation.
+// Injected LLMClient implementations are not retried automatically.
+// With Observer, Run emits started before facade request conversion or a nil
+// engine check, and finished on every normal return after public error mapping
+// and resource release. The event RunID matches a successful result. Without
+// Observer, result timing retains its existing convention; observed wall
+// duration can be slightly longer because it spans the facade boundary.
 //
 // Run can return every error category documented by [Engine.Prepare], plus
-// ErrCapacityExceeded and ErrLLMGenerate. An engine admission rejection is
+// ErrCapacityExceeded and ErrLLMGenerate. Built-in HTTP execution failures
+// additionally match ErrTransport; malformed successful responses and nil
+// successful injected responses additionally match ErrMalformedResponse.
+// An engine admission rejection is
 // discoverable as [CapacityError] and still matches ErrCapacityExceeded. It
 // occurs before artifacts, schemas, rendering, or model generation because the
 // selected backend's admission capacity is full; it does not match
 // ErrInvalidRequest or ErrLLMGenerate. A built-in OpenAI-compatible non-2xx
-// response is discoverable as [GenerationError]. Errors from injected clients
-// remain available through errors.Is. Cancellation while waiting for
+// response is discoverable as [GenerationError]. Eligible-status exhaustion
+// with a positive budget is discoverable as [RetryExhaustedError]. Injected
+// errors remain available through errors.Is. Cancellation while waiting for
 // model-generation capacity matches both ErrLLMGenerate and the context error.
 // Cancellation otherwise follows the active collaborator's documented
 // behavior. A nil Engine returns ErrInvalidConfig. Run returns no partial
 // result on error.
 func (e *Engine) Run(ctx context.Context, req RunRequest) (*RunResult, error) {
-	if e == nil || e.runner == nil {
+	ctx = observation.WithoutExecutionState(ctx)
+	if req.Observer == nil {
+		return e.run(ctx, req)
+	}
+	recorder, idErr := observation.New(func(event domain.ExecutionEvent) { req.Observer(fromDomainExecutionEvent(event)) })
+	recorder.Start()
+	if idErr != nil {
+		err := fmt.Errorf("failed to create run id: %w", idErr)
+		terminal := observedFinish(nil, err)
+		recorder.Finish(observation.FinishData{Outcome: terminal.Outcome, Failure: terminal.Failure})
+		return nil, err
+	}
+	result, err := e.run(observation.WithRecorder(ctx, recorder), req)
+	terminal := observedFinish(result, err)
+	recorder.Finish(observation.FinishData{Outcome: terminal.Outcome, Failure: terminal.Failure,
+		StatusCode: terminal.StatusCode, RetriesExhausted: terminal.RetriesExhausted})
+	return result, err
+}
+
+func (e *Engine) run(ctx context.Context, req RunRequest) (*RunResult, error) {
+	if e == nil || e.internal == nil {
 		return nil, fmt.Errorf("%w: engine is nil", ErrInvalidConfig)
 	}
 
@@ -666,7 +510,7 @@ func (e *Engine) Run(ctx context.Context, req RunRequest) (*RunResult, error) {
 		return nil, fmt.Errorf("%w: %v", ErrInvalidRequest, err)
 	}
 
-	result, err := e.runner.Run(ctx, domainReq)
+	result, err := e.internal.Run(ctx, domainReq)
 	if err != nil {
 		return nil, mapPublicError(err)
 	}
@@ -676,6 +520,11 @@ func (e *Engine) Run(ctx context.Context, req RunRequest) (*RunResult, error) {
 // RunPrepared atomically claims and executes a handle created by
 // [Engine.PrepareExecution].
 //
+// The originating Engine and namespace views sharing its underlying owner may
+// claim the handle. Execution uses its captured selections without applying the
+// invoking view's namespaces. An Engine constructed separately by [NewEngine]
+// is a foreign owner.
+//
 // A valid owning-Engine invocation consumes the handle's one attempt before
 // credential revalidation, backend admission, generation, or validation.
 // Cancellation, capacity rejection, generation failure, operational
@@ -686,19 +535,30 @@ func (e *Engine) Run(ctx context.Context, req RunRequest) (*RunResult, error) {
 //
 // The supplied context governs this execution attempt independently of the
 // preparation context. It covers credential revalidation, admission,
-// generation, validation, and any internal repair. Result timing begins after
-// the claim and excludes preparation and consumer-held delay.
+// generation, validation, and any bounded output repair. Result timing begins
+// after the claim and excludes preparation and consumer-held delay.
 //
 // RunPrepared can return ErrInvalidRequest, ErrAPIKeyEnvMissing,
 // ErrCapacityExceeded, ErrLLMGenerate, or ErrValidation as applicable while
 // preserving documented collaborator and context identities. An engine
 // admission rejection is discoverable as [CapacityError] and still matches
 // ErrCapacityExceeded. A built-in OpenAI-compatible non-2xx response is
-// discoverable as [GenerationError]. A completed content-validation rejection
-// is returned in RunResult, not as an operational error. An operational error
-// returns no partial RunResult.
+// discoverable as [GenerationError]. Built-in HTTP execution failures match
+// ErrTransport; malformed successful responses and nil successful injected
+// responses match ErrMalformedResponse. Positive-budget eligible-status
+// exhaustion is discoverable as [RetryExhaustedError]. A completed
+// content-validation rejection, including repair exhaustion, is returned in
+// RunResult, not as an operational
+// error. RunPrepared uses the retry choice frozen during PrepareExecution. An
+// operational error returns no partial RunResult. A retained Observer begins
+// only after a successful atomic claim, with identities from the frozen
+// snapshot, and receives a terminal event on normal return after public error
+// mapping. Invalid, foreign, discarded, or reused handles emit no events.
+// Callbacks run outside the handle and capacity locks; panics propagate and
+// do not promise a terminal event.
 func (e *Engine) RunPrepared(ctx context.Context, prepared *PreparedExecution) (*RunResult, error) {
-	if e == nil || e.runner == nil {
+	ctx = observation.WithoutExecutionState(ctx)
+	if e == nil || e.internal == nil {
 		return nil, fmt.Errorf("%w: engine is nil", ErrInvalidConfig)
 	}
 
@@ -706,9 +566,21 @@ func (e *Engine) RunPrepared(ctx context.Context, prepared *PreparedExecution) (
 	if prepared != nil {
 		internal = prepared.internal
 	}
-	result, err := e.runner.RunPrepared(ctx, internal)
+	var recorder *observation.Recorder
+	result, err := e.internal.RunPrepared(observation.WithClaimSink(ctx, &recorder), internal)
 	if err != nil {
-		return nil, mapPublicError(err)
+		mapped := mapPublicError(err)
+		if recorder != nil {
+			terminal := observedFinish(nil, mapped)
+			recorder.Finish(observation.FinishData{Outcome: terminal.Outcome, Failure: terminal.Failure,
+				StatusCode: terminal.StatusCode, RetriesExhausted: terminal.RetriesExhausted})
+		}
+		return nil, mapped
 	}
-	return fromDomainRunResult(result), nil
+	publicResult := fromDomainRunResult(result)
+	if recorder != nil {
+		terminal := observedFinish(publicResult, nil)
+		recorder.Finish(observation.FinishData{Outcome: terminal.Outcome})
+	}
+	return publicResult, nil
 }

errors.go

@@ -5,11 +5,12 @@ import (
 	"fmt"
 	"strings"
 
-	"gitea.maximumdirect.net/eric/promptkit/internal/capacity"
-	"gitea.maximumdirect.net/eric/promptkit/internal/llm"
-	"gitea.maximumdirect.net/eric/promptkit/internal/profile"
-	"gitea.maximumdirect.net/eric/promptkit/internal/promptdef"
-	"gitea.maximumdirect.net/eric/promptkit/internal/usecase"
+	"go.maximumdirect.net/promptkit/internal/capacity"
+	"go.maximumdirect.net/promptkit/internal/generation"
+	"go.maximumdirect.net/promptkit/internal/llm"
+	"go.maximumdirect.net/promptkit/internal/profile"
+	"go.maximumdirect.net/promptkit/internal/promptdef"
+	"go.maximumdirect.net/promptkit/internal/usecase"
 )
 
 func mapPublicError(err error) error {
@@ -21,6 +22,17 @@ func mapPublicError(err error) error {
 		strings.TrimSpace(internalCapacityError.BackendID) != "" {
 		return &CapacityError{BackendID: internalCapacityError.BackendID}
 	}
+	var internalExhausted *generation.ExhaustedError
+	if errors.As(err, &internalExhausted) && internalExhausted != nil {
+		mapped := newRetryExhaustedError(internalExhausted.Attempts(), mapPublicError(internalExhausted.Unwrap()))
+		if errors.Is(err, usecase.ErrValidation) && !errors.Is(mapped, ErrValidation) {
+			return fmt.Errorf("%w: %w", ErrValidation, mapped)
+		}
+		if publicErr := publicErrorFor(err); publicErr != nil && !errors.Is(mapped, publicErr) {
+			return fmt.Errorf("%w: %w", publicErr, mapped)
+		}
+		return mapped
+	}
 	publicErr := publicErrorFor(err)
 	var providerHTTPError *llm.ProviderHTTPError
 	if errors.As(err, &providerHTTPError) && providerHTTPError != nil {
@@ -38,7 +50,14 @@ func mapPublicError(err error) error {
 	if publicErr == nil {
 		return err
 	}
-	return fmt.Errorf("%w: %w", publicErr, err)
+	switch {
+	case errors.Is(err, llm.ErrTransport):
+		return fmt.Errorf("%w: %w: %w", publicErr, ErrTransport, err)
+	case errors.Is(err, llm.ErrMalformedResponse):
+		return fmt.Errorf("%w: %w: %w", publicErr, ErrMalformedResponse, err)
+	default:
+		return fmt.Errorf("%w: %w", publicErr, err)
+	}
 }
 
 func publicErrorFor(err error) error {
@@ -53,10 +72,6 @@ func publicErrorFor(err error) error {
 		return ErrPromptLoad
 	case errors.Is(err, usecase.ErrProfileLoad):
 		return ErrProfileLoad
-	case errors.Is(err, promptdef.ErrInvalidYAML), errors.Is(err, promptdef.ErrInvalidPromptDefinition):
-		return ErrPromptLoad
-	case isProfileLoadCause(err):
-		return ErrProfileLoad
 	case errors.Is(err, usecase.ErrAPIKeyEnvMissing):
 		return errors.Join(ErrInvalidRequest, ErrAPIKeyEnvMissing)
 	case errors.Is(err, capacity.ErrCapacityExceeded):
@@ -75,9 +90,3 @@ func publicErrorFor(err error) error {
 		return nil
 	}
 }
-
-func isProfileLoadCause(err error) bool {
-	return errors.Is(err, profile.ErrInvalidYAML) ||
-		errors.Is(err, profile.ErrInvalidProfile) ||
-		errors.Is(err, profile.ErrRawAPIKeyNotAllowed)
-}

errors_internal_test.go

@@ -6,8 +6,8 @@ import (
 	"fmt"
 	"testing"
 
-	"gitea.maximumdirect.net/eric/promptkit/internal/llm"
-	"gitea.maximumdirect.net/eric/promptkit/internal/usecase"
+	"go.maximumdirect.net/promptkit/internal/llm"
+	"go.maximumdirect.net/promptkit/internal/usecase"
 )
 
 func TestMapPublicErrorPreservesGenerationCancellation(t *testing.T) {
@@ -73,3 +73,10 @@ func TestMapPublicErrorPreservesValidationAroundGenerationError(t *testing.T) {
 		t.Fatalf("mapped error exposes internal ProviderHTTPError: %v", err)
 	}
 }
+
+func TestMapPublicErrorDoesNotTreatRequestEncodingAsTransport(t *testing.T) {
+	err := mapPublicError(fmt.Errorf("%w: %w", usecase.ErrLLMGenerate, llm.ErrRequestFailed))
+	if !errors.Is(err, ErrLLMGenerate) || errors.Is(err, ErrTransport) {
+		t.Fatalf("request encoding category = %v", err)
+	}
+}

examples/go-library/observe/main.go

@@ -0,0 +1,173 @@
+package main
+
+import (
+	"context"
+	"embed"
+	"encoding/json"
+	"fmt"
+	"io"
+	"net/http"
+	"os"
+	"strings"
+
+	"go.maximumdirect.net/configkit"
+	"go.maximumdirect.net/promptkit"
+)
+
+//go:embed prompts/*.yaml schemas/*.json
+var assets embed.FS
+
+type scriptedTransport struct {
+	proposalCalls int
+}
+
+func (t *scriptedTransport) RoundTrip(req *http.Request) (*http.Response, error) {
+	if err := req.Context().Err(); err != nil {
+		return nil, err
+	}
+	var request struct {
+		Model string `json:"model"`
+	}
+	if err := json.NewDecoder(req.Body).Decode(&request); err != nil {
+		return nil, err
+	}
+	switch request.Model {
+	case "proposal-model":
+		t.proposalCalls++
+		if t.proposalCalls == 1 {
+			return httpResponse(http.StatusServiceUnavailable, `{"error":{}}`), nil
+		}
+		return modelResponse(`{"title":"Queue cleanup","risk":"low"}`, 10, 8), nil
+	case "assessment-model":
+		return modelResponse(`{"recommendation":"accept","confidence":0.9}`, 7, 5), nil
+	default:
+		return nil, fmt.Errorf("unexpected model in offline transport")
+	}
+}
+
+func modelResponse(content string, promptTokens, completionTokens int) *http.Response {
+	body, _ := json.Marshal(map[string]any{
+		"choices": []any{map[string]any{"message": map[string]string{"content": content}}},
+		"usage":   map[string]int{"prompt_tokens": promptTokens, "completion_tokens": completionTokens},
+	})
+	return httpResponse(http.StatusOK, string(body))
+}
+
+func httpResponse(status int, body string) *http.Response {
+	return &http.Response{StatusCode: status, Header: make(http.Header), Body: io.NopCloser(strings.NewReader(body))}
+}
+
+type eventLog struct{ events []promptkit.ExecutionEvent }
+
+func (l *eventLog) record(event promptkit.ExecutionEvent) { l.events = append(l.events, event) }
+
+func (l *eventLog) terminal() (promptkit.ExecutionEvent, error) {
+	if len(l.events) == 0 || l.events[len(l.events)-1].Kind != promptkit.ExecutionFinished {
+		return promptkit.ExecutionEvent{}, fmt.Errorf("execution did not finish")
+	}
+	return l.events[len(l.events)-1], nil
+}
+
+type summary struct {
+	ProposalAttempts           uint64 `json:"proposal_attempts"`
+	ProposalTransientRetries   uint64 `json:"proposal_transient_retries"`
+	ProposalUsageComplete      bool   `json:"proposal_usage_complete"`
+	AssessmentAttempts         uint64 `json:"assessment_attempts"`
+	AssessmentTransientRetries uint64 `json:"assessment_transient_retries"`
+	AssessmentOutcome          string `json:"assessment_outcome"`
+}
+
+func run() (summary, error) {
+	transport := &scriptedTransport{}
+	inputs, err := promptkit.FilesystemInputs("example prompts", assets, "prompts")
+	if err != nil {
+		return summary{}, err
+	}
+	backendID, proposalModel, assessmentModel := "offline-backend", "proposal-model", "assessment-model"
+	backend := promptkit.LocalBackend("https://example.invalid/v1", 1)
+	backend.ID = backendID
+	inputs = append(inputs,
+		configkit.Input{Source: promptkit.BackendSource("example backend", []promptkit.BackendConfig{
+			backend,
+		})},
+		configkit.Input{Source: promptkit.ProfileSource("example profiles", []promptkit.ProfileConfig{
+			{ID: "proposal-profile", Backend: &backendID, Model: &proposalModel},
+			{ID: "assessment-profile", Backend: &backendID, Model: &assessmentModel},
+		})},
+	)
+	engine, err := promptkit.NewEngine(promptkit.Config{
+		Sources:    promptkit.ConfigurationSources{Underlays: []configkit.Underlay{{Inputs: inputs}}},
+		HTTPClient: &http.Client{Transport: transport},
+	}, promptkit.WithBundledOpenRouter(false), promptkit.WithSchemaFS(assets, "schemas")) // Keep the example's configuration self-contained.
+	if err != nil {
+		return summary{}, err
+	}
+
+	var proposalLog eventLog
+	proposal, err := engine.Run(context.Background(), promptkit.RunRequest{
+		PromptID: "example.proposal", Inputs: map[string]promptkit.ArtifactRef{"idea": promptkit.Inline("Reduce queue cleanup toil.")}, Observer: proposalLog.record,
+	})
+	if err != nil {
+		return summary{}, err
+	}
+	if !proposal.Validation.IsValid {
+		return summary{}, fmt.Errorf("proposal failed structural validation")
+	}
+	proposalTerminal, err := proposalLog.terminal()
+	if err != nil {
+		return summary{}, err
+	}
+	var assessmentLog eventLog
+	assessment, err := engine.Run(context.Background(), promptkit.RunRequest{
+		PromptID: "example.assessment", Inputs: map[string]promptkit.ArtifactRef{"proposal": promptkit.Inline(proposal.RawOutput)}, Observer: assessmentLog.record,
+	})
+	if err != nil {
+		return summary{}, err
+	}
+	if !assessment.Validation.IsValid {
+		return summary{}, fmt.Errorf("assessment failed structural validation")
+	}
+	assessmentTerminal, err := assessmentLog.terminal()
+	if err != nil {
+		return summary{}, err
+	}
+	var proposalValue struct {
+		Risk string `json:"risk"`
+	}
+	var assessmentValue struct {
+		Recommendation string `json:"recommendation"`
+	}
+	if err := json.Unmarshal([]byte(proposal.RawOutput), &proposalValue); err != nil {
+		return summary{}, err
+	}
+	if err := json.Unmarshal([]byte(assessment.RawOutput), &assessmentValue); err != nil {
+		return summary{}, err
+	}
+	decision := "rejected"
+	if proposalValue.Risk == "low" && assessmentValue.Recommendation == "accept" {
+		decision = "accepted"
+	}
+	return summary{
+		ProposalAttempts:           proposalTerminal.Attempts,
+		ProposalTransientRetries:   proposalTerminal.TransientRetries,
+		ProposalUsageComplete:      proposalTerminal.Usage.PromptTokens.Complete,
+		AssessmentAttempts:         assessmentTerminal.Attempts,
+		AssessmentTransientRetries: assessmentTerminal.TransientRetries,
+		AssessmentOutcome:          decision,
+	}, nil
+}
+
+func main() {
+	if err := execute(os.Stdout); err != nil {
+		fmt.Fprintln(os.Stderr, err)
+		os.Exit(1)
+	}
+}
+
+func execute(out io.Writer) error {
+	value, err := run()
+	if err != nil {
+		return err
+	}
+	return json.NewEncoder(out).Encode(value)
+}

examples/go-library/observe/main_test.go

@@ -0,0 +1,30 @@
+package main
+
+import (
+	"bytes"
+	"encoding/json"
+	"testing"
+)
+
+func TestObserveExample(t *testing.T) {
+	var output bytes.Buffer
+	if err := execute(&output); err != nil {
+		t.Fatal(err)
+	}
+	var got struct {
+		ProposalAttempts   *int   `json:"proposal_attempts"`
+		ProposalRetries    *int   `json:"proposal_transient_retries"`
+		UsageComplete      *bool  `json:"proposal_usage_complete"`
+		AssessmentAttempts *int   `json:"assessment_attempts"`
+		AssessmentRetries  *int   `json:"assessment_transient_retries"`
+		Outcome            string `json:"assessment_outcome"`
+	}
+	if err := json.Unmarshal(output.Bytes(), &got); err != nil {
+		t.Fatal(err)
+	}
+	if got.ProposalAttempts == nil || *got.ProposalAttempts != 2 || got.ProposalRetries == nil || *got.ProposalRetries != 1 ||
+		got.UsageComplete == nil || *got.UsageComplete || got.AssessmentAttempts == nil || *got.AssessmentAttempts != 1 ||
+		got.AssessmentRetries == nil || *got.AssessmentRetries != 0 || got.Outcome != "accepted" {
+		t.Fatalf("unexpected observation summary: %+v", got)
+	}
+}

examples/go-library/observe/prompts/assessment.yaml

@@ -0,0 +1,17 @@
+prompts:
+  - id: example.assessment
+    default_profile: assessment-profile
+    description: Assess a structured proposal independently of schema validity.
+    inputs:
+      - name: proposal
+        required: true
+        content_type: application/json
+    messages:
+      - role: system
+        content: Return a concise assessment as JSON.
+      - role: user
+        content: '{{input "proposal"}}'
+    output:
+      format: json
+      validation_mode: json_schema
+      schema_path: assessment.schema.json

examples/go-library/observe/prompts/proposal.yaml

@@ -0,0 +1,17 @@
+prompts:
+  - id: example.proposal
+    default_profile: proposal-profile
+    description: Propose a bounded improvement from a synthetic idea.
+    inputs:
+      - name: idea
+        required: true
+        content_type: text/plain
+    messages:
+      - role: system
+        content: Return a concise proposal as JSON.
+      - role: user
+        content: '{{input "idea"}}'
+    output:
+      format: json
+      validation_mode: json_schema
+      schema_path: proposal.schema.json

examples/go-library/observe/schemas/assessment.schema.json

@@ -0,0 +1,10 @@
+{
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "type": "object",
+  "required": ["recommendation", "confidence"],
+  "properties": {
+    "recommendation": {"type": "string"},
+    "confidence": {"type": "number"}
+  },
+  "additionalProperties": false
+}

examples/go-library/observe/schemas/proposal.schema.json

@@ -0,0 +1,10 @@
+{
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "type": "object",
+  "required": ["title", "risk"],
+  "properties": {
+    "title": {"type": "string"},
+    "risk": {"type": "string"}
+  },
+  "additionalProperties": false
+}

examples/go-library/prepare/collections/review/collection.yaml

@@ -0,0 +1,17 @@
+backends:
+  - id: offline
+    endpoint: https://review.example.invalid/v1
+profiles:
+  - id: audit
+    backend: review/offline
+    model: review-model
+prompts:
+  - id: check
+    default_profile: review/audit
+    description: Inspect a review item without contacting a model provider.
+    messages:
+      - role: user
+        content: Review the supplied synthetic example.
+    output:
+      format: text
+      validation_mode: none

examples/go-library/prepare/collections/support/collection.yaml

@@ -0,0 +1,10 @@
+version: "2"
+config:
+  imports:
+    - path: fragments/resources.yaml
+      required: true
+    - path: fragments/prepare.yaml
+      destination: [prompts]
+      placement: sequence-item
+      required: true
+release_version: "1.4.0"

examples/go-library/prepare/collections/support/fragments/prepare.yaml

@@ -1,6 +1,5 @@
-id: example.prepare
-version: "1.0.0"
-default_profile: offline-example
+id: prepare
+default_profile: support/fast
 description: Prepare a prompt without contacting a model provider.
 inputs:
   - name: note
@@ -10,7 +9,7 @@ messages:
   - role: system
     content: Summarize the note in one sentence.
   - role: user
-    content: '{{input "note"}}'
+    content_file: ../messages/note.md
 output:
   format: text
   validation_mode: basic

examples/go-library/prepare/collections/support/fragments/resources.yaml

@@ -0,0 +1,11 @@
+backends:
+  - id: offline
+    endpoint: https://example.invalid/v1
+profiles:
+  - id: base
+    backend: support/offline
+    model: offline-model
+    timeout_seconds: 90
+  - id: fast
+    extends: {id: base}
+    timeout_seconds: 60

examples/go-library/prepare/collections/support/messages/note.md

@@ -0,0 +1 @@
+{{input "note"}}

examples/go-library/prepare/collections/support/operator-overrides.yaml

@@ -0,0 +1,5 @@
+# A later input replaces the profile with an inherited variant.
+profiles:
+  - id: fast
+    extends: {id: base}
+    timeout_seconds: 30

examples/go-library/prepare/main.go

@@ -4,57 +4,162 @@ import (
 	"context"
 	"encoding/json"
 	"fmt"
+	"io"
 	"os"
+	"path/filepath"
 
-	"gitea.maximumdirect.net/eric/promptkit"
+	"go.maximumdirect.net/configkit"
+	"go.maximumdirect.net/promptkit"
 )
 
 type summary struct {
-	PromptID        string `json:"prompt_id"`
-	PromptVersion   string `json:"prompt_version"`
-	SelectedProfile string `json:"selected_profile"`
-	Model           string `json:"model"`
-	MessageCount    int    `json:"message_count"`
+	PromptID        string              `json:"prompt_id"`
+	SelectedProfile string              `json:"selected_profile"`
+	Model           string              `json:"model"`
+	MessageCount    int                 `json:"message_count"`
+	ProfileTimeout  int                 `json:"profile_timeout_seconds"`
+	Namespaces      []string            `json:"namespaces"`
+	Collections     []collectionSummary `json:"collections"`
+	ScopedPrompts   []string            `json:"scoped_prompts"`
+	ReviewPrompts   []string            `json:"review_prompts"`
+}
+
+type collectionSummary struct {
+	Namespace      string  `json:"namespace"`
+	ReleaseVersion *string `json:"release_version"`
 }
 
 func main() {
-	engine, err := promptkit.NewEngine(
-		promptkit.Config{},
-		promptkit.WithPromptFile("examples/go-library/prepare/prompt.yaml"),
-		promptkit.WithProfiles(promptkit.Profile{
-			ID:       "offline-example",
-			Endpoint: "https://example.invalid/v1",
-			Model:    "offline-model",
-		}),
+	if err := run(context.Background(), "examples/go-library/prepare", os.Stdout); err != nil {
+		fmt.Fprintln(os.Stderr, err)
+		os.Exit(1)
+	}
+}
+
+func run(ctx context.Context, exampleDir string, out io.Writer) error {
+	support, err := promptkit.CollectionInput(
+		"support",
+		configkit.File(filepath.Join(exampleDir, "collections", "support", "collection.yaml")),
 	)
 	if err != nil {
-		exit(err)
+		return err
+	}
+	review, err := promptkit.CollectionInput(
+		"review",
+		configkit.File(filepath.Join(exampleDir, "collections", "review", "collection.yaml")),
+	)
+	if err != nil {
+		return err
+	}
+	override, err := promptkit.CollectionInput(
+		"support",
+		configkit.File(filepath.Join(exampleDir, "collections", "support", "operator-overrides.yaml")),
+	)
+	if err != nil {
+		return err
+	}
+	engine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{
+		Underlays: []configkit.Underlay{
+			{Inputs: []configkit.Input{support, review}},
+			{Inputs: []configkit.Input{override}},
+		},
+	}}, promptkit.WithBundledOpenRouter(false)) // Keep this example's inventory caller-owned.
+	if err != nil {
+		return err
+	}
+	scoped, err := engine.WithNamespaces(promptkit.Namespaces{
+		Prompt: "support", Profile: "support", Backend: "support",
+	})
+	if err != nil {
+		return err
+	}
+	view, err := engine.InspectConfiguration()
+	if err != nil {
+		return err
+	}
+	if _, found := view.Explain(configkit.Keys("collections", "support", "profiles").Index(1).Key("timeout_seconds")); !found {
+		return fmt.Errorf("collection profile override was not found")
+	}
+	requirements, err := scoped.InspectRequirements(ctx, promptkit.RequirementInspectionRequest{
+		PromptID: "prepare",
+	})
+	if err != nil {
+		return err
+	}
+	if !requirements.Complete || len(requirements.Inputs) != 1 ||
+		requirements.Inputs[0].Name != "note" ||
+		requirements.Inputs[0].Kind != promptkit.RequirementRequired {
+		return fmt.Errorf("required note input was not discovered before preparation")
 	}
 
-	prepared, err := engine.Prepare(context.Background(), promptkit.RunRequest{
-		PromptID: "example.prepare",
+	request := promptkit.RunRequest{
+		PromptID:  "prepare",
+		ProfileID: "fast",
 		Inputs: map[string]promptkit.ArtifactRef{
 			"note": promptkit.Inline("Ada finished the migration review."),
 		},
+	}
+	if err := scoped.ValidateRequest(ctx, request); err != nil {
+		return err
+	}
+	prepared, err := scoped.Prepare(ctx, request)
+	if err != nil {
+		return err
+	}
+	profile, err := scoped.InspectProfile(ctx, "fast")
+	if err != nil {
+		return err
+	}
+	if profile.EffectiveModelParams.TimeoutSeconds != 30 {
+		return fmt.Errorf(
+			"later profile override was not applied: timeout is %d seconds",
+			profile.EffectiveModelParams.TimeoutSeconds,
+		)
+	}
+	namespaces, err := scoped.ListNamespaces(ctx)
+	if err != nil {
+		return err
+	}
+	collections, err := scoped.ListCollections(ctx)
+	if err != nil {
+		return err
+	}
+	scopedPrompts, err := scoped.ListPrompts(ctx)
+	if err != nil {
+		return err
+	}
+	reviewPrompts, err := scoped.ListPrompts(ctx, promptkit.NamespaceFilter{
+		Mode: promptkit.NamespaceNamed, Name: "review",
 	})
 	if err != nil {
-		exit(err)
+		return err
+	}
+	scopedPromptIDs := make([]string, len(scopedPrompts))
+	for index := range scopedPrompts {
+		scopedPromptIDs[index] = scopedPrompts[index].ID
+	}
+	reviewPromptIDs := make([]string, len(reviewPrompts))
+	for index := range reviewPrompts {
+		reviewPromptIDs[index] = reviewPrompts[index].ID
+	}
+	collectionSummaries := make([]collectionSummary, len(collections))
+	for index := range collections {
+		collectionSummaries[index] = collectionSummary{
+			Namespace: collections[index].Namespace, ReleaseVersion: collections[index].ReleaseVersion,
+		}
 	}
 
-	encoder := json.NewEncoder(os.Stdout)
+	encoder := json.NewEncoder(out)
 	encoder.SetIndent("", "  ")
-	if err := encoder.Encode(summary{
+	return encoder.Encode(summary{
 		PromptID:        prepared.PromptID,
-		PromptVersion:   prepared.PromptVersion,
 		SelectedProfile: prepared.SelectedProfileID,
 		Model:           prepared.EffectiveModelParams.Model,
 		MessageCount:    len(prepared.Messages),
-	}); err != nil {
-		exit(err)
-	}
-}
-
-func exit(err error) {
-	fmt.Fprintln(os.Stderr, err)
-	os.Exit(1)
+		ProfileTimeout:  profile.EffectiveModelParams.TimeoutSeconds,
+		Namespaces:      namespaces,
+		Collections:     collectionSummaries,
+		ScopedPrompts:   scopedPromptIDs,
+		ReviewPrompts:   reviewPromptIDs,
+	})
 }

examples/go-library/prepare/main_test.go

@@ -0,0 +1,45 @@
+package main
+
+import (
+	"bytes"
+	"encoding/json"
+	"reflect"
+	"testing"
+)
+
+func TestPrepareExample(t *testing.T) {
+	var output bytes.Buffer
+	if err := run(t.Context(), ".", &output); err != nil {
+		t.Fatal(err)
+	}
+	var got struct {
+		PromptID      string                       `json:"prompt_id"`
+		Profile       string                       `json:"selected_profile"`
+		Model         string                       `json:"model"`
+		MessageCount  int                          `json:"message_count"`
+		Timeout       int                          `json:"profile_timeout_seconds"`
+		Namespaces    []string                     `json:"namespaces"`
+		Collections   []map[string]json.RawMessage `json:"collections"`
+		ScopedPrompts []string                     `json:"scoped_prompts"`
+		ReviewPrompts []string                     `json:"review_prompts"`
+	}
+	if err := json.Unmarshal(output.Bytes(), &got); err != nil {
+		t.Fatal(err)
+	}
+	if got.PromptID != "support/prepare" || got.Profile != "support/fast" || got.Model != "offline-model" ||
+		got.MessageCount <= 0 || got.Timeout != 30 || !reflect.DeepEqual(got.Namespaces, []string{"review", "support"}) ||
+		!reflect.DeepEqual(got.ScopedPrompts, []string{"support/prepare"}) || !reflect.DeepEqual(got.ReviewPrompts, []string{"review/check"}) || len(got.Collections) != 2 {
+		t.Fatalf("unexpected preparation summary: %+v", got)
+	}
+	for index, want := range []struct{ namespace, release string }{{"review", ""}, {"support", "1.4.0"}} {
+		var namespace string
+		if err := json.Unmarshal(got.Collections[index]["namespace"], &namespace); err != nil || namespace != want.namespace {
+			t.Fatalf("collection namespace = %q, %v", namespace, err)
+		}
+		value, present := got.Collections[index]["release_version"]
+		var release *string
+		if !present || json.Unmarshal(value, &release) != nil || (index == 0 && release != nil) || (index == 1 && (release == nil || *release != want.release)) {
+			t.Fatalf("collection release = %s, present %t; want %q", value, present, want.release)
+		}
+	}
+}

examples/go-library/run/main.go

@@ -4,9 +4,12 @@ import (
 	"context"
 	"encoding/json"
 	"fmt"
+	"io"
 	"os"
+	"path/filepath"
 
-	"gitea.maximumdirect.net/eric/promptkit"
+	"go.maximumdirect.net/configkit"
+	"go.maximumdirect.net/promptkit"
 )
 
 type deterministicClient struct{}
@@ -38,44 +41,41 @@ type summary struct {
 }
 
 func main() {
-	engine, err := promptkit.NewEngine(
-		promptkit.Config{},
-		promptkit.WithPromptFile("examples/go-library/run/prompt.yaml"),
-		promptkit.WithProfiles(promptkit.Profile{
-			ID:       "offline-example",
-			Endpoint: "https://example.invalid/v1",
-			Model:    "offline-model",
-		}),
-		promptkit.WithLLMClient(deterministicClient{}),
-	)
+	if err := run(context.Background(), "examples/go-library/run", os.Stdout); err != nil {
+		fmt.Fprintln(os.Stderr, err)
+		os.Exit(1)
+	}
+}
+
+func run(ctx context.Context, exampleDir string, out io.Writer) error {
+	endpoint, model := "https://example.invalid/v1", "offline-model"
+	engine, err := promptkit.NewEngine(promptkit.Config{Sources: promptkit.ConfigurationSources{
+		Underlays: []configkit.Underlay{{Inputs: []configkit.Input{{Source: promptkit.ProfileSource("example profile", []promptkit.ProfileConfig{
+			{ID: "offline-example", Endpoint: &endpoint, Model: &model},
+		})}}}},
+		Main: &configkit.Input{Source: configkit.File(filepath.Join(exampleDir, "prompt.yaml"))},
+	}}, promptkit.WithBundledOpenRouter(false), promptkit.WithLLMClient(deterministicClient{})) // Use only this example's offline configuration.
 	if err != nil {
-		exit(err)
+		return err
 	}
 
-	result, err := engine.Run(context.Background(), promptkit.RunRequest{
+	result, err := engine.Run(ctx, promptkit.RunRequest{
 		PromptID: "example.run",
 		Inputs: map[string]promptkit.ArtifactRef{
 			"note": promptkit.Inline("Ada finished the migration review."),
 		},
 	})
 	if err != nil {
-		exit(err)
+		return err
 	}
 
-	encoder := json.NewEncoder(os.Stdout)
+	encoder := json.NewEncoder(out)
 	encoder.SetIndent("", "  ")
-	if err := encoder.Encode(summary{
+	return encoder.Encode(summary{
 		Output:           result.RawOutput,
 		ValidationStatus: result.Validation.Status,
 		IsValid:          result.Validation.IsValid,
 		Model:            result.ModelName,
 		TotalTokens:      result.Usage.TotalTokens,
-	}); err != nil {
-		exit(err)
-	}
-}
-
-func exit(err error) {
-	fmt.Fprintln(os.Stderr, err)
-	os.Exit(1)
+	})
 }

examples/go-library/run/main_test.go

@@ -0,0 +1,28 @@
+package main
+
+import (
+	"bytes"
+	"encoding/json"
+	"testing"
+)
+
+func TestRunExample(t *testing.T) {
+	var output bytes.Buffer
+	if err := run(t.Context(), ".", &output); err != nil {
+		t.Fatal(err)
+	}
+	var got struct {
+		Output string `json:"output"`
+		Valid  *bool  `json:"is_valid"`
+		Status string `json:"validation_status"`
+		Model  string `json:"model"`
+		Tokens *int   `json:"total_tokens"`
+	}
+	if err := json.Unmarshal(output.Bytes(), &got); err != nil {
+		t.Fatal(err)
+	}
+	if got.Output != "Ada finished the migration review." || got.Valid == nil || !*got.Valid ||
+		got.Status != "passed" || got.Model != "offline-model" || got.Tokens == nil || *got.Tokens != 18 {
+		t.Fatalf("unexpected execution summary: %+v", got)
+	}
+}

examples/go-library/run/prompt.yaml

@@ -1,16 +1,16 @@
-id: example.run
-version: "1.0.0"
-default_profile: offline-example
-description: Run a prompt with a deterministic injected model client.
-inputs:
-  - name: note
-    required: true
-    content_type: text/plain
-messages:
-  - role: system
-    content: Summarize the note in one sentence.
-  - role: user
-    content: '{{input "note"}}'
-output:
-  format: text
-  validation_mode: basic
+prompts:
+  - id: example.run
+    default_profile: offline-example
+    description: Run a prompt with a deterministic injected model client.
+    inputs:
+      - name: note
+        required: true
+        content_type: text/plain
+    messages:
+      - role: system
+        content: Summarize the note in one sentence.
+      - role: user
+        content: '{{input "note"}}'
+    output:
+      format: text
+      validation_mode: basic