Tighten pr-reviewer autofix boundaries#60
Conversation
|
Caution Review failedPull request was closed or merged during review 📝 WalkthroughWalkthroughThe review agent's scope is narrowed from general auto-fixing to mechanical-only fixes. Persona, harness prompt, documentation, and tests are updated to enforce and verify that the agent auto-applies only lint/format/typo/import-order changes, leaving logic, safety-critical code, lifecycle paths, and tests as reviewer comments requiring human ownership. ChangesReview Agent Scope Narrowing
Estimated code review effort🎯 2 (Simple) | ⏱️ ~8 minutes Possibly related issues
Possibly related PRs
Suggested labels
Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
There was a problem hiding this comment.
Code Review
This pull request updates the PR reviewer agent to adopt a more conservative behavior. Instead of proactively fixing all issues, the agent is now restricted to auto-applying only mechanical, non-semantic fixes (such as linting, formatting, typos, and import ordering). It explicitly forbids auto-editing semantic or safety-critical logic, changing safety defaults, modifying lifecycle/termination code, or adding/modifying tests to make its own changes pass. Unit tests have been added to verify these new prompt constraints. I have no feedback to provide on these changes.
Important
The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.
|
ℹ️ pr-reviewer: review only — no file changes were applied to the PR (nothing to commit after review). The notes below are advisory and were not pushed. No PR breakage found in the current checkout, so I left the PR files unchanged. Addressed comments
Advisory NotesNone. Local validation run:
I am not printing |
|
No source changes were needed. I reviewed the PR diff against the current checkout and found the prompt/docs/persona/test updates consistent with the PR’s mechanical-only auto-fix goal. Addressed comments
Advisory NotesNone. Local validation:
I am not printing |
Summary
Tightens the
pr-reviewerpersona soagent-relay-code[bot]no longer treats broad logic changes as safe auto-pushes.Pinned persona/workflow:
review/persona.tsreview/agent.tsreview/README.mdrg "apply pr-reviewer fixes|pr-reviewer|agent-relay-code|autofix|auto.?fix|push|no-agent-relay-review"confirmed the active workflow is thereview/persona. Its harness prompt is the cloud-commit boundary: cloud pushes whatever the harness leaves in the PR worktree.Guardrails Added
timeout/pending/throw/undefined becomingacked/true/{}no-agent-relay-review.Evidence For This Tightening
factory-sdk#266 /e76232a: changedrecord.fixtureFiles ? ...toObject.prototype.hasOwnProperty.call(record, 'fixtureFiles') ? ...infleet.tsloadConfig. That truthiness-to-presence swap re-selected FakeMount forfixtureFiles: null/{}and reintroduced the pulled:0 inert-binary bug V2FIX-10b fixed. This was a fail-open regression.factory-sdk#264 /c77a088: added a per-issue prune of#dispatchFailureReaperHandoffson dispatch success. That was wrong scope/timing and orphaned in-flight crash protection; the safe version belongs in 11c as per-pair/on-confirmed-dead..test.tsassertions for their new behavior (+53/+29/+17 lines observed), making CI green on the bot's own change. Self-justifying tests defeat the gate.confirmWritefromtimeouttoacked, another fail-closed to fail-open regression.Verification
npm run test:reviewpassed: 31/31.npx agentworkforce persona compile review/persona.tspassed.git diff --checkpassed.Repo-wide
npm run typecheckandnpm teststill fail on existing unrelated baseline TypeScript errors across multiple personas/runtime event typings before this patch's tests run.