auto-review-loop-minimax

Automate multi-round external research reviews by triggering an autonomous loop where MiniMax critiques submissions, implements fixes, and re-reviews until acceptance criteria are met. This reduces manual back-and-forth and improves turnaround for research artifacts.

State persistence saves logs and progress to REVIEW_STATE.json and AUTO_REVIEW.md files, allowing the multi-round research review process to resume seamlessly after interruptions without losing tracked evaluation rounds.

Yes, you need an API key read from environment variables to use MiniMax for iterative research critique. The Skill integrates with MiniMax via MCP tool, with a curl fallback for flexible backend execution.

The best way to apply iterative feedback loops is using an autonomous review system that performs review, fix implementation, and re-evaluation until a positive assessment or maximum rounds are reached.

Yes, you can use this autonomous review loop for code submissions and experiments. It applies rigorous external critique and iterative improvement to research papers, code, and experiments requiring multi-round evaluation.

When a research artifact fails to meet acceptance criteria, the loop continues implementing fixes and re-reviewing until a positive assessment is achieved or the defined MAX_ROUNDS limit is reached, terminating the autonomous process.