What problem does it solve?
Researchers often have a visible problem but a fuzzy technical route, leading to bloated proposals, contribution sprawl, or drift away from the original bottleneck. This Skill turns a vague research direction into a problem-anchored, implementation-ready method plan through structured multi-round external review.
Core Features & Use Cases
- Problem Anchor Freezing: Extracts an immutable bottom-line problem, constraints, and success conditions that every revision round must preserve, with explicit drift detection.
- Iterative GPT-5.4 Review Loop: Sends proposals to a Codex-based reviewer across up to 5 rounds, scoring 7 dimensions (problem fidelity, method specificity, contribution quality, frontier leverage, feasibility, validation focus, venue readiness) 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 from the last completed phase within 24 hours.
- Use Case: A PhD student has a rough idea for improving long-context reasoning in LLMs. The skill scans local papers, identifies the technical gap, drafts a focused method proposal with minimal claim-driven experiments, and iteratively refines it against reviewer feedback until it is top-venue ready, then hands off to an experiment-planning skill.
Quick Start
Ask the assistant to refine your research approach by describing your problem and rough idea, for example: refine my approach to reducing hallucination in retrieval-augmented generation using a lightweight verifier.