Skip to content
Get packs

Auto Review Loop (MiniMax)

Runs up to 4 rounds of automated research review and fixes using the MiniMax API.

Source Download skill
Repository
wanshuiyin/Auto-claude-code-research-in-sleep
Stars
18,482
Licence
MIT
Files
1 file

What it does

Gives the agent a repeatable routine for improving a research project: it sends the work to an external AI reviewer (MiniMax-M3), reads the score and critique, makes fixes or runs experiments, and asks for another review. It stops when the reviewer rates the work as ready or almost ready, or after 4 rounds. It saves progress to a state file so an interrupted run can resume.

Great for

  • Researchers preparing a paper who want a second opinion on draft claims and experiments before submission
  • Teams that cannot use the Codex MCP reviewer and prefer MiniMax as the external reviewer

Skip it if

  • Projects with confidential or unpublished data you do not want sent to a third-party API, because review prompts include research context and the reviewer is an external service
  • Anyone expecting a human-quality peer review; the reviewer is a language model and its scores are advisory

Try asking

  • “Run an auto review loop minimax on my paper draft in this repo”
  • “minimax review of the current experiment results and the claims in the introduction”
  • “Use MiniMax to review my submission and fix the top weaknesses for up to four rounds”
Step-by-step guide: how to use Auto Review Loop (MiniMax)Install, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 1 original files, the MIT licence and an attribution note.

    auto-review-loop-minimax.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip auto-review-loop-minimax.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip auto-review-loop-minimax.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 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

  • Give the skill a clear scope in the argument, such as a specific section or claim set, so the reviewer gets focused context
  • Do not wrap it in /loop, /schedule or cron, as the SKILL.md warns; it already loops internally and an external timer discards accumulated context
  • Check review-stage/AUTO_REVIEW.md after each run, since the full raw reviewer responses are logged there

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.