auto-review-loop-minimax

Automating conference-level paper revision involves an autonomous loop that sends your work to an external reviewer, implements the minimum necessary fixes, and re-submits for re-evaluation until quality criteria are met or a maximum round limit is reached.

Yes, automating research review with the MiniMax API uses an MCP tool for reviewer calls with a curl fallback. It drives experiment automation by repeatedly assessing claims, methods, and results, then applying fixes within an iterative loop.

State recovery in an autonomous review loop persists the current review stage as a JSON file. This allows the automated paper revision process to safely resume after context compaction or interrupted runs without losing the iterative critique progress.

To run an iterative critique loop for ICML or NeurIPS level papers, trigger the autonomous review process with your topic. The loop collects a ranked list of weaknesses, implements minimum fixes, and documents each round in an auto review markdown file.

Yes, the automated paper revision loop works without MCP by falling back to a curl-based API call for the MiniMax reviewer. This ensures the autonomous multi-round review and experiment automation process remains functional across different environments.