auto-review-loop-minimax

To automate multi-round external review for research submissions, trigger the autonomous review loop to iteratively assess and improve your work. The process applies configurable MAX_ROUNDS, positive-threshold criteria, and state persistence to track progress across cycles.

State persistence in an autonomous review pipeline tracks progress across iterative review cycles using a REVIEW_STATE.json file. It ensures the system maintains a complete audit trail, preserving the status and data of each round until threshold criteria are met.

To start an autonomous review loop using the MiniMax API, command "auto review loop minimax". This initiates a round-based external review process that uses primary MCP-based review when available, with a curl fallback to the MiniMax API.

Yes, you can configure the maximum rounds and threshold criteria for research reviews by setting MAX_ROUNDS and positive-threshold parameters. This allows you to tailor the iterative review process to meet the specific rigor requirements of different projects.

Yes, the automated review loop generates and updates an AUTO_REVIEW.md file and a REVIEW_STATE.json file. This automatic generation ensures structured logging and keeps a complete audit trail of each round's assessment and improvements.

A limitation of using a MiniMax powered review loop is its dependency on external API availability, requiring either MCP integration or a curl fallback. Additionally, the review process is bound by your configured MAX_ROUNDS, which halts the loop if criteria are unmet.