Auto Review Loop (Generic LLM)
Repeatedly has an external LLM review, critique and revise a research project until it's judged ready.
- Stars
- 18,482
- Licence
- MIT
- Files
- 1 file
What it does
Once installed, the agent runs a review-fix-re-review cycle on a research project: it sends the project's claims, methods and results to an external LLM through any OpenAI-compatible API, asks for a score and verdict, plans fixes, and repeats for up to 4 rounds or until the reviewer says the work is ready. Each round is logged to a review document and progress is saved to a state file so the loop can resume after interruption.
Great for
- Researchers preparing a paper who want a second, automated critical read before submission
- Students who want rigorous, ranked feedback on drafts and experiment plans
- Teams who want to compare reviews from different LLM providers
Skip it if
- Anyone expecting a human-quality peer review: the scores come from another language model and can be overconfident or wrong
- Sensitive or unpublished work sent to a third-party API, since project content leaves your machine and is governed by that provider's terms
- Projects where the agent should not edit files autonomously, because the skill is allowed to run shell commands and write or edit files
Try asking
- “Run an auto review loop llm on my paper draft in this repo”
- “llm review of my experiment results and suggest the minimum fixes”
- “Start the auto review loop llm for the topic of sparse attention and keep a round log”
Install in 3 steps
- 1
Download the skill
The ZIP contains the 1 original files, the MIT licence and an attribution note.
auto-review-loop-llm.zip - 2
Add it to your agent
Claude Code (all projects)
mkdir -p ~/.claude/skills && unzip auto-review-loop-llm.zip -d ~/.claude/skills/Claude Code / Agent SDK (this project only)
mkdir -p .claude/skills && unzip auto-review-loop-llm.zip -d .claude/skills/Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.
- 3
Ask for the task
No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.
Get better results
- Set LLM_API_KEY through the MCP server env block rather than pasting keys into prompts, and review the model and base URL before running.
- Do not wrap this skill in /loop or /schedule; it already loops internally and an external timer discards accumulated context.
- Check the AUTO_REVIEW.md log after each run, and verify fixes yourself rather than accepting the reviewer's verdict as final.
What's inside
- SKILL.md
Redistributed unmodified from wanshuiyin/Auto-claude-code-research-in-sleep under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.