review-reference-maintenance

To prevent stale code review references from causing incorrect AI feedback, you need periodic reference maintenance to audit and reclassify entries by status, review age, and applicability to the current codebase. This ensures active references remain accurate and reusable.

The best way to triage outdated review guidance is to evaluate active entries first based on review age and codebase applicability, then assign candidate or deprecated status based on concrete evidence of context clarity and reusability strength.

To audit a references.yaml file for team-lead review guidance, inspect each entry to evaluate reusability and context clarity, update statuses to active, candidate, or deprecated, and generate a structured maintenance result with counts and human-review flags.

To remove personal names and overly contextual phrasing from team workflow references, harden the references by distilling lessons to eliminate emotion and context-dependent details, ensuring the guidance stays broadly applicable for future AI code reviews.

You should deprecate a review reference when it lacks applicability evidence, context clarity, or reusability strength for the current design direction, moving it from active to deprecated status while flagging entries that require human review.

No specific dependencies are required to perform review-reference maintenance. The process operates directly on your team-lead references configuration to evaluate reusability criteria and output a structured maintenance summary with status counts and human-review flags.