What problem does it solve? Researchers often have a clear problem but only a vague technical approach, leading to bloated proposals, contribution sprawl, or methods that drift away from the original bottleneck. This Skill turns a fuzzy research direction into a focused, implementation-ready method plan through structured, multi-round external critique. ## Core Features & Use Cases - Problem Anchor Freezing: Extracts and locks an immutable problem statement (bottleneck, non-goals, constraints, success criteria) that is reused in every refinement round to prevent drift. - Iterative External Review Loop: Sends proposals to an external LLM reviewer (llm-chat MCP) for strict scoring across 7 dimensions, then revises until the score reaches 9 or the round cap is hit. - Checkpoint Recovery: Persists state to REFINE_STATE.json after each phase so interrupted sessions can resume within 24 hours without losing completed work. - Use Case: A PhD student has a rough idea for improving mechanistic interpretability probing. They run the skill, which scans local papers, drafts a focused method proposal, gets it critiqued by an external reviewer, and iterates until the plan is sharp enough to hand off to experiment planning. ## Quick Start Ask the assistant to refine your research approach by describing your problem and current vague method idea, for example by invoking the research-refine skill with your problem statement and draft approach.
