Complete pre-1.0 release hardening
This commit is contained in:
@@ -897,6 +897,26 @@ This ledger should not replace existing reports. It should provide a flattened r
|
||||
|
||||
Finish 1.0 by turning the new architecture into a documented, tested, stable release candidate.
|
||||
|
||||
## Implementation status (2026-05-13)
|
||||
|
||||
This workstream is now implemented:
|
||||
- architecture, contract, configuration, output-schema, structured-LLM, validator, prompt, diagnostics, and subprocess docs are aligned to the current runtime;
|
||||
- a practical release checklist document exists (`docs/release-checklist.md`);
|
||||
- curated release fixtures exist under `internal/cli/testdata/release` and run under `go test ./...` using fake structured LLM responses only;
|
||||
- release fixture coverage includes:
|
||||
- must-apply and must-not-apply expectations;
|
||||
- protected-term survival;
|
||||
- default module sequence and stable instance naming;
|
||||
- report schema metadata;
|
||||
- diagnostics artifact references;
|
||||
- output schema checks (`bare-segments`, `audita-v1`);
|
||||
- prompt/structured-schema diagnostics metadata presence;
|
||||
- stable validator keys;
|
||||
- utilization diagnostics and correction-ledger artifacts;
|
||||
- idempotence-oriented second-pass no-op expectations with deterministic fake responses.
|
||||
|
||||
Deferred work remains deferred (no filesystem prompt overrides, no user-configurable validator chains, no arbitrary user-supplied output schemas, no resume/start-at/stop-after execution, no diff/check/propose-only modes, no generated transcript descriptions enabled by default, no interactive review UI, no UI/server wrapper, and no provider-specific benchmarking harness).
|
||||
|
||||
## Documentation updates
|
||||
|
||||
Update or add:
|
||||
|
||||
Reference in New Issue
Block a user