auto-review-loop-llm

Automate iterative research review by cycling through LLM critique, fix implementation, and re-assessment until defined completion thresholds are met. This loop continuously improves research artifacts by incorporating suggestions and re-submitting for review.

Yes, automated research review supports any OpenAI-compatible API. You can configure the LLM integration via an MCP server or direct curl calls to facilitate the continuous improvement process.

An iterative LLM review loop is an autonomous process running multiple rounds of critique and improvement on research artifacts. You need it to get continuous feedback on a paper draft until it meets submission standards for top venues.

Recover interrupted review progress using state persistence. The process saves progress to a REVIEW_STATE.json file, allowing you to resume autonomous review cycles without losing your critique and fix implementation data.

Yes, the autonomous review loop supports configuration via environment variables or an MCP server. This allows you to define LLM settings and completion thresholds directly within your development environment.

Avoid using an automated LLM review loop if your research artifact requires subjective human oversight rather than defined completion thresholds. The autonomous cycles are designed for continuous mechanical improvement, not nuanced editorial judgment.