Auto Review Loop Llm

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

Runs an autonomous review loop where an external LLM (via OpenAI-compatible API) critiques your research, you implement fixes, then it re-reviews up to four rounds until it scores 6/10 or higher with a "ready" or "almost" verdict. Works with DeepSeek, OpenAI, MiniMax, Kimi, and other compatible providers through the llm-chat MCP server. The skill already loops internally, so the documentation explicitly warns against wrapping it in /loop or cron schedules since that would waste tokens by dropping accumulated context on each tick. Every round gets logged to AUTO_REVIEW.md with the full raw reviewer response preserved in a collapsible detail block, and state persists to REVIEW_STATE.json for crash recovery. Useful when you want honest external feedback without manually running the review-fix cycle yourself.

skills.sh installs: 525
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-llm --agent claude-code

Installs into .claude/skills of the current project.

Files

Auto Review Loop (Generic LLM): Autonomous Research Improvement

🔒 Do not wrap this skill in /loop, /schedule, or CronCreate. Like /auto-review-loop, it already loops internally (review → fix → re-review), feeding each round's prior-round summary into the next review prompt (the backend is a stateless per-round API/MCP call, not a shared thread). An external timer re-enters from the top each tick, dropping that accumulated context and firing the verdict on wall-clock time instead of on artifact change — zero new signal, full token cost. Schedule the external wait that precedes it, not the verdict. See shared-references/external-cadence.md.

Autonomously iterate: review → implement fixes → re-review, until the external reviewer gives a positive assessment or MAX_ROUNDS is reached.