What problem does it solve? Getting rigorous, unbiased feedback on research ideas, papers, and experimental results is hard without access to senior reviewers. This Skill connects your project to an external LLM reviewer through the llm-chat MCP server, delivering NeurIPS/ICML-level critical reviews with iterative dialogue until claims, narrative, and experiment plans converge. ## Core Features & Use Cases - Multi-Round External Review: Sends comprehensive research context to an external LLM reviewer and iterates through rounds of criticism, rebuttal, and refinement. - Strict Reviewer Configuration: Resolves LLM_MODEL, LLM_BASE_URL, and LLM_API_KEY from project .mcp.json, user settings, or shell environment with a hard-fail rule when unconfigured. - Actionable Deliverables: Produces mock conference reviews, minimal experiment packages, claims matrices, paper outlines, and a self-contained review document saved to the project. - Use Case: Before submitting a mechanistic interpretability paper, run this Skill to get a brutal mock review, identify missing experiments, and generate a results-to-claims matrix for each possible experimental outcome. ## Quick Start Ask the assistant to review my research using the research-review skill with the topic of my current project.
