Document the new configuration and processing workflow
This commit is contained in:
@@ -77,8 +77,8 @@ hidden global state.
|
||||
- Validate corrections before final mutation, including local proposal checks,
|
||||
decision cardinality, and apply-time safety. Provider-side schema enforcement
|
||||
does not replace these checks.
|
||||
- Keep LLM concurrency bounded and context-aware through PromptKit's engine-local
|
||||
per-backend admission. Proposal and validation calls selecting the same backend
|
||||
- Use PromptKit's context-aware, engine-local per-backend admission for
|
||||
LLM concurrency. Proposal and validation calls selecting the same backend
|
||||
share that capacity, including retries and repairs. Preserve deterministic
|
||||
collection order. Recoverable model-operation failures stay local to their
|
||||
proposal response or validation batch; fatal invariant/application failure
|
||||
@@ -120,13 +120,19 @@ responsibilities and should remain separate. Implementation details belong in
|
||||
|
||||
## Dependencies
|
||||
|
||||
Keep Audita dependency-light and prefer the Go standard library for CLI parsing,
|
||||
filesystems, synchronization, and tests. ConfigKit owns configuration loading,
|
||||
presence, provenance, and secret resolution; PromptKit owns prompt/profile
|
||||
loading, OpenAI-compatible transport, retries, repairs, and backend admission.
|
||||
Audita owns domain meaning, role selection, and deterministic processing.
|
||||
Keep Audita dependency-light and prefer the Go standard library where no adopted
|
||||
library owns the boundary. ConfigKit owns configuration composition, presence,
|
||||
recorded resolution, and provenance. PromptKit owns prompt, profile, backend,
|
||||
credential, transport, retry, repair, and admission behavior. CommandKit and its
|
||||
integrations own command routing and their standard inspection and request
|
||||
surfaces. Audita owns domain meaning, role selection, transcript processing,
|
||||
and presentation of its processing results.
|
||||
|
||||
Do not introduce broad CLI, dependency-injection, workflow, logging, or plugin
|
||||
Pass resolved typed configuration to preflight and processing. Preflight checks
|
||||
execution readiness before output preparation or model work; the output session
|
||||
owns resources, cleanup, and publication. The glossary and readiness decision is
|
||||
recorded in [ADR 0002](../adr/0002-glossary-configuration-and-preflight.md).
|
||||
Do not introduce broad dependency-injection, workflow, logging, or plugin
|
||||
frameworks without a concrete implemented need and focused tests.
|
||||
|
||||
## Testing and Documentation
|
||||
|
||||
@@ -184,6 +184,15 @@ 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.
|
||||
|
||||
## Audita's Selected Candidate Checks
|
||||
|
||||
Audita uses ReleaseKit v0.9.0 with race-enabled offline tests, full Go vet,
|
||||
formatting, Markdown links, module distribution checks, and the required target
|
||||
builds. Race checks address shared processing and client state; vet and format
|
||||
checks catch common defects and mechanical drift; link and module checks protect
|
||||
the distributed source tree. The [runbook](../release.md) specifies the selected
|
||||
toolchain, native commands, and asynchronous recovery.
|
||||
|
||||
## Publication Guards
|
||||
|
||||
Automate the checks needed to publish the intended source under a fresh version:
|
||||
|
||||
Reference in New Issue
Block a user