A housing-portal PR keeps getting the same Claude Code review comments on every re-run. What should re-reviews do?
Select an answer to reveal the explanation.
Short Explanation
Nobody likes sticky notes that get re-stuck on the same crack every morning. Carry prior findings forward and ask only for new or still-open issues.
Full Explanation
Housing-portal PRs that re-run Claude Code review on every push must manage comment hygiene. Effective re-reviews include prior findings in context and instruct the model to report only new issues or still-open defects. That pattern preserves institutional memory of what was already flagged, avoids duplicate sticky comments on the same lines, and keeps developer attention on remaining work instead of noise.
Wiping prior findings so every comment can be posted again fails because it recreates the same noise storm each commit, trains developers to ignore the bot, and obscures whether anything was actually fixed. Posting duplicate comments on every commit to maximize noise fails by design: volume is not signal, and municipal teams will mute or disable review when trust collapses. Disabling review after the first pass regardless of remaining defects fails because later commits can introduce new bugs or reopen fixed issues; a one-and-done policy abandons continuous quality gates.
Exam caveat: carrying prior findings forward reduces duplicates; it does not auto-verify that a "fixed" item is truly resolved—CI should still re-check open items against the latest diff. Operational check: on re-review, pass the previous finding list into the prompt, require output that marks each prior item resolved/still-open/new, and configure the comment poster to update or skip rather than spam identical inline notes.