Auto Review Loop

wanshuiyin/auto-claude-code-research-in-sleep

Loops a code review backend until your work passes or hits four rounds. You tell it what to review, it calls out to Codex or a manual reviewer, parses the critique, implements fixes, and resubmits. Stops when score hits 6/10 and verdict is "ready" or "almost". The skill writes everything to a review-stage directory and persists state so it can survive context compaction mid-loop. You can run it fully autonomous or pause after each round to vet the fixes yourself. It defaults to Codex at medium difficulty, but you can crank it to nightmare mode where the reviewer gets direct repo access and actively tries to poke holes in your claims. Good for iterating on research artifacts or any work where you want aggressive external feedback without manually shuttling files back and forth.

skills.sh installs: 602
GitHub stars: 18.9K
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add wanshuiyin/auto-claude-code-research-in-sleep --skill auto-review-loop --agent claude-code

Installs into .claude/skills of the current project.

Files

Auto Review Loop: Autonomous Research Improvement

🔒 Do not wrap this skill in /loop, /schedule, or CronCreate. It already loops internally (review → fix → re-review) and the reviewer carries round-to-round memory in one threadId (codex-reply). An external timer re-enters from the top each tick — fresh threadId, reviewer memory reset — firing the verdict on wall-clock time instead of on artifact change: zero new signal, full token cost. If you want to schedule something, schedule the external wait that precedes it (experiments done → then run this once). See shared-references/external-cadence.md.

Autonomously iterate: review → implement fixes → re-review, until an independent reviewer gives a policy-approved positive assessment or MAX_ROUNDS is reached.