What problem does it solve? Research papers and ML projects need repeated external critique before submission, but manually soliciting reviewer feedback is slow and inconsistent. This Skill runs an autonomous loop that sends your work to an external LLM reviewer, parses the score and weaknesses, implements fixes, and re-reviews until the work is judged ready or a round limit is reached. ## Core Features & Use Cases - Autonomous Review Loop: Iterates review → fix → re-review for up to 4 rounds, stopping early when the score reaches 6/10 or the verdict indicates readiness. - Provider-Agnostic LLM Access: Works with any OpenAI-compatible API (OpenAI, DeepSeek, Kimi, GLM, SiliconFlow, and more) via an llm-chat MCP server or a curl fallback. - State Persistence & Recovery: Writes REVIEW_STATE.json after every round and maintains a cumulative AUTO_REVIEW.md log with verbatim reviewer responses. - Use Case: A researcher preparing a NeurIPS submission triggers the loop; the Skill obtains a brutally honest score, implements the minimum fixes for each ranked weakness, and re-submits until the reviewer says the paper is ready. ## Quick Start Ask the AI to run the auto review loop llm on the current research project and iterate until the external reviewer accepts it.
