auto-review-loop-llm

It coordinates an external ML reviewer and an OpenAI-compatible LLM to autonomously read context, identify weaknesses, and propose fixes across multiple rounds until criteria are met or limits are reached.

Multi-round research artifact improvement involves continuous review, fix, and re-review cycles. This skill applies configurable round limits and persists review state to ensure documentation and traceability for academic or applied ML projects.

Yes, you can use any OpenAI-compatible LLM provider. The skill supports configurable MCP-based LLM interaction and includes a curl-based fallback to connect to multiple providers for iterative research reviews.

You can configure round limits to control the maximum number of autonomous review and fix cycles. The skill enforces these limits and mandates documentation of all rounds to maintain traceability and repeatability.

Use an automated LLM review loop for research pipelines, model documentation updates, and submission readiness checks for papers or reports to iteratively improve work without manual intervention.

Limitations include dependency on an external reviewer and a max round limit to prevent infinite cycles. The process requires persistent review state documentation, which may increase storage overhead for large ML projects.