auto-review-loop-minimax

To automate iterative external reviews, this Skill orchestrates a multi-round review loop using the MiniMax API to critique and guide research improvements until reaching a positive assessment or maximum rounds. It autonomously reviews, implements fixes, and re-reviews the work.

An autonomous review loop is a workflow that repeatedly critiques and refines research. This Skill ensures state persistence and recovery by saving progress to REVIEW_STATE.json and AUTO_REVIEW.md, providing resilience and auditability across review rounds.

You can run multi-round research critiques by triggering the autonomous workflow, which uses a primary MCP tool with a curl fallback to the MiniMax API. It processes project artifacts to generate a fresh external assessment and guide iterative fixes.

Yes, you can use curl as a fallback for the MiniMax API when the primary MCP tool is unavailable. This flexible API option ensures the autonomous review loop continues to obtain external critique without interruption.

You need round-based workflow orchestration when you want external review to systematically guide improvements across multiple rounds. This approach satisfies requirements for state persistence, API fallback, and reproducible, auditable research refinement.

The primary limitation of autonomous research critique loops is the maximum round threshold, which caps the number of iterations. If a positive assessment is not reached by MAX_ROUNDS, the loop terminates, requiring manual intervention for further improvements.