Adopt PromptKit v0.23.0 conversations and execution accounting

This commit is contained in:
2026-11-15 01:25:55 +00:00
parent d28b560265
commit d98d419b36
25 changed files with 689 additions and 924 deletions

View File

@@ -48,6 +48,14 @@ artifacts can still contain sensitive transcript and prompt content. Execution
position and optional operator ID are carried separately from generated safe
artifact path coordinates.
Execution diagnostics include a native-group invocation summary and per-call
summaries. Terminal coverage is separate from usage completeness; unknown usage
is distinct from known zero. Overlapping summaries must not be added together.
The collector snapshots after work joins and retains domain labels and stable
presentation order. Prepared evidence comes from `OnPrepared`, so preparation or
admission failures may have no request snapshot. Coordinator execution failures
still finalize the attempted logical call and its available evidence.
## Key Tests
Useful tests: `internal/core/reporting/*_test.go`,

View File

@@ -9,27 +9,44 @@ it never replaces the requested engine.
Each resolved pipeline entry receives role-bound clients with its selected
profiles and effective budgets. Clients share engine-local backend admission.
The adapter uses `PrepareExecution` and `RunPrepared` on the same handle, passes
The adapter uses ordinary `Run` with `OperationTimeout` and `OnPrepared`, passes
Audita's per-call schema and repair budget, and decodes only validated results.
The operation budget includes preparation, admission, retries, and structural
repair. `ErrOperationTimeout` remains recoverable even when wrapped by a
preparation error. Parent cancellation remains distinct. The prepared notification
provides the actual request snapshot; failures before notification have no snapshot.
PromptKit owns backend and credential selection. The adapter uses captured
credential source names for artifact redaction without inspecting unrelated
environment values. See the [integration contract](../integrations/openai-compatible-llm.md)
for failure behavior and the [configuration reference](../config.md) for sources.
When correction repair is enabled, the proposal call retains its prompt,
effective variables, prepared messages, and full final response for section-local
replay. The adapter renders the repair definition with the selected profile,
appends only that branch’s history, and checks the original prepared prefix
before execution. The complete conversation is sent to the selected provider;
provider rejection ends that branch as unavailable. The requested repair
definition is retained in call attribution alongside native base-prompt metadata.
When correction repair is enabled, the proposal call captures a native immutable
PromptKit exchange. Each actionable correction starts an independent
`Engine.Coordinate` branch from that exchange. PromptKit assembles continuation
history, owns the additional-turn budget, and delivers the last permitted response
for assessment. Follow-ups select the repair prompt, profile, and output schema
directly; Audita does not rerender the original proposal or assemble replay messages.
An opaque exchange in the structured-output contract keeps provider state outside
modules and validators. The adapter shares request construction, notifications,
observers, and result/failure finalization between ordinary and coordinated calls.
Audita owns actionable rejection, revision decoding, withdrawal, current previews,
and full revalidation. Accepted siblings remain independent. Requesting more work
at the budget boundary produces exhaustion without allocating another logical call.
A canceled, undelivered response supplies diagnostic evidence only; it cannot enter
the approved set. Repair disabled in configuration bypasses coordination entirely.
The runner supplements each repair instruction with its current proposal preview
as described in the [module guide](modules.md#failure-and-outcome-behavior).
The bundled prompts separate durable system principles from the initial user
assignment so later repair instructions can allow acceptable minor cleanup;
see [prompt instruction ownership](prompts.md).
The observation collector tracks logical calls and execution events. Ordinary
The observation collector tracks logical calls and execution events, and enrolls
selected runs in a native `ExecutionGroup`. Proposal, repair, and nested validator
clients explicitly share the invocation observer; history reuse adds no extra run.
Native summaries deduplicate terminal totals by run ID and retain terminal coverage
separately from usage presence and completeness. Logical calls and run membership
can differ, especially on failures before execution. Ordinary
reports use compact safe outcomes; request and response evidence is optional
debug material. The runner preserves ordered pipeline traversal and deterministic
section collection while canceling and joining peers on fatal errors.

View File

@@ -20,7 +20,9 @@ formats; [configuration](../config.md) owns settings; the
Recoverable failed batches mark their candidates unavailable; successful
survivors continue through all required validators.
6. Actionable rejections can enter independent section-local repair branches.
Each revision is mapped to the original section and revalidated from the
PromptKit coordinates each branch from the captured proposal exchange, owning
its history and additional-turn limit. Each revision, including the final
permitted response, is mapped to the original section and revalidated from the
start. Withdrawal, exhaustion, and recoverable failures finish the branch
without affecting independent approvals.
7. Only fully approved proposals reach application. Corrections apply in