What problem does it solve? Getting objective, rigorous feedback on research work before submission is slow and depends on scarce human reviewers. This Skill runs an autonomous review loop that scores your work, identifies weaknesses, implements fixes, and re-reviews until the work meets a quality threshold. ## Core Features & Use Cases - Autonomous Review Loop: Iterates review → fix → re-review for up to 4 rounds, stopping when the score reaches 6/10 with a "ready" or "almost" verdict. - Provider-Agnostic LLM Backend: Works with any OpenAI-compatible API (OpenAI, DeepSeek, Kimi, GLM, SiliconFlow, and more) via the llm-chat MCP server or direct curl calls. - State Persistence & Recovery: Saves round state to REVIEW_STATE.json and logs cumulative reviews to AUTO_REVIEW.md, enabling recovery after interruptions. - Use Case: A researcher preparing a NeurIPS submission runs the loop overnight; each round feeds the previous review summary into the next prompt, fixes are applied, and the loop stops when the external reviewer judges the paper ready. ## Quick Start Run an auto review loop with the LLM reviewer on my current research draft and iterate until it is ready for submission.
