sysadminandClaude Opus 4.8
f20e0fc338
Reviewer runs that observe a validation failure and later rerun to green must report the failure and its investigation; a later pass never erases the earlier failure: - New reviewer_validation_failure_history.py verifier: reports must carry a Validation failure history section naming every observed failure with command, failing test or error, suspected cause, reproduced status, baseline master status, and PR-caused evidence; rerun-pass claims must explain what changed between runs, whether environmental state (such as stale /tmp files) was cleaned, and why the pass is trustworthy; an unexplained transient failure cannot be reported as a bare pass and must use the status 'passed after transient failure investigation' (exported as TRANSIENT_INVESTIGATION_STATUS). - Re-export from review_proofs and wire into build_final_report as a downgrade check. - Review-merge final-report schema gains the Validation failure history template section; contract test locks the markers and an export test locks the re-export. - 12 tests covering the issue's scenarios: failing-then-passing rerun, baseline-equivalent failure, /tmp state contamination, unexplained transient failure, and a report omitting an earlier failure. Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
3.4 KiB
Review-merge controller handoff schema
Task mode: review-merge-pr
End every review/merge run with a section titled exactly Controller Handoff.
Use this canonical field set. Do not omit fields — use none or
not verified in this session where appropriate.
Do not use legacy fields: Pinned reviewed head, Scratch worktree used,
Workspace mutations, Mutations: None (when mutations occurred).
## Controller Handoff
- Task:
- Repo:
- Role:
- Identity:
- Active profile:
- Runtime context:
- Selected PR:
- Linked issue:
- Eligibility class:
- Queue ordering policy:
- Inventory pagination proof:
- Earlier PRs skipped:
- Candidate head SHA:
- Reviewed head SHA:
- Target branch:
- Target branch SHA:
- Already-landed gate:
- Author-safety result:
- Prior request-changes state:
- Review worktree used:
- Review worktree path:
- Review worktree inside branches:
- Review worktree HEAD state:
- Review worktree dirty before validation:
- Review worktree dirty after validation:
- Baseline worktree used:
- Baseline worktree path:
- Files reviewed:
- Validation:
- Official validation integrity status:
- Terminal review mutation:
- Review decision:
- Merge preflight:
- Merge result:
- Linked issue status:
- Main checkout branch:
- Main checkout dirty state:
- Main checkout updated:
- File edits by reviewer:
- Worktree/index mutations:
- Git ref mutations:
- MCP/Gitea mutations:
- Review mutations:
- Merge mutations:
- Cleanup mutations:
- External-state mutations:
- Read-only diagnostics:
- Blockers:
- Current status:
- Safe next action:
- Safety statement:
Already-landed handoff overrides
When eligibility class is ALREADY_LANDED_RECONCILE_REQUIRED:
- Reviewed head SHA:
none - Review worktree used:
false - Review worktree path:
none - Review decision:
none - Merge result:
none
Identity format: username / profile (not personal email unless required — #305).
Queue-status-only runs (no selected PR)
When the run inventories the queue but selects no PR for review:
- Selected PR:
none - Already-landed gate, Author-safety result, Merge preflight:
not applicableornot run— neverpassed - Review worktree detail fields:
not applicableornonewhen Review worktree used isfalse - Blockers must not be
noneif the narrative says all open PRs are conflicted, blocked, or unverified - Inventory pagination proof must cite final-page metadata, not default page-size assumptions
Verifier: review_proofs.assess_queue_status_report().
Narrative final report and controller handoff must agree on eligibility class, candidate/reviewed head SHA, mutation state, worktree usage, review decision, terminal review mutation, merge result, and linked issue status.
Validation failure history (#396)
Every validation failure observed during the run must appear here, even if a later rerun passed. A later pass never erases the earlier failure.
For each observed failure report:
- Command:
- Failing test or error:
- Suspected cause:
- Reproduced:
- Baseline master:
- PR-caused evidence:
If a later rerun passed, also report what changed between runs, whether any
environmental state (for example stale /tmp files) was cleaned, and why the
pass is trustworthy.
If the cause is unknown, the validation status must be
passed after transient failure investigation — never a bare passed.
Verifier: review_proofs.assess_validation_failure_history_report().