feat(memory): learn evidence-backed recovery policies (#1022)#1109
Conversation
Add a bounded recovery policy learner so repeated evidence-backed recoveries can bias future candidate ranking without bypassing safety gates or executing actions.\n\nConstraint: Learned recovery policy must be advisory only and must not store sensitive page content or credentials.\nRejected: A separate autonomous recovery memory/executor | duplicates PatternLearner direction and risks unsafe automatic action.\nConfidence: high\nScope-risk: narrow\nDirective: Learned policies may rank candidates but must never override risk gates or opt-in execution controls.\nTested: npm test -- --runTestsByPath tests/recovery/policy-learner.test.ts tests/recovery/candidate-ranker.test.ts; npm run build\nNot-tested: Two-run live OpenChrome policy promotion transcript.
|
Warning You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again! |
Qodo reviews are paused for this user.Troubleshooting steps vary by plan Learn more → On a Teams plan? Using GitHub Enterprise Server, GitLab Self-Managed, or Bitbucket Data Center? |
|
Codex usage limits have been reached for code reviews. Please check with the admins of this repo to increase the limits by adding credits. |
Merge rationale (stack consolidation)Intent. Closes #1022 — adds a Why this is correct.
CI. Targets the recovery-candidate-ranking feature branch; CI workflow only runs on |
1 participant
Progress / Review status
Auto-refreshed 2026-05-13 — owner comments cleaned up to reduce review noise.
feat/1022-recovery-policy-learning→feat/1018-recovery-candidate-ranking8565816— Learn evidence-backed recovery ranking policiesOwner comment cleanup: 0 issue + 0 inline review comments deleted. Outstanding feedback from automated/external reviewers above is unchanged.
Summary\n- Adds RecoveryPolicyLearner for evidence-backed recovery outcome promotion and persistence.\n- Lets candidate ranking accept promoted policies as a bounded score boost.\n- Keeps learned policies advisory-only and unable to bypass safety/risk gates.\n\nCloses #1022.\n\nStacked on #1088 because it biases the recovery candidate ranker.\n\n## Validation\n- npm test -- --runTestsByPath tests/recovery/policy-learner.test.ts tests/recovery/candidate-ranker.test.ts\n- npm run build\n\n## Live OpenChrome verification\n- Not run in this PR session; requires two-run stale-ref fixture transcript after the stacked foundations land.\n\n## Safety\n- Only evidence-backed outcomes are recorded.\n- Ambiguous outcomes do not promote.\n- Policies only bias ranking; they do not auto-execute actions and do not bypass risk gates.