research-refine
Turn a vague research direction into a problem-anchored, elegant, frontier-aware, implementation-oriented method plan via iterative GPT-5.4 review. Use when the user says \"refine my approach\", \"帮我细化方案\", \"decompose this problem\", \"打磨idea\", \"refine research plan\", \"细化研究方案\", or wants a concrete research method that stays simple, focused, and top-venue ready instead of a vague or overbuilt idea.
npx skills add majiayu000/claude-skill-registry --skill research-refine-wanshuiyin-auto-claude-code-res-2 --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
What it does
The skill guides a user from a visible research problem to a concrete, implementation-oriented method plan anchored to a fixed problem. It enforces an iterative review process (Phase 0 to Phase 4) that culminates in a full anchored proposal and a minimal, validation-focused plan. It uses a structured workflow of problem freezing, grounding-gap identification, and a single-dominant-contribution thesis, with explicit stages for analysis, refinement, and documentation. It prescribes tooling and roles (Claude, Codex, Claude Code) and outputs to manage: initial proposal, multi-round reviews, and a final proposal with a history log. It also defines evaluation via an external reviewer round and a scoring rubric across several dimensions, followed by revision instructions and handoff mechanics.
How it works
- Phase 0: Freeze the Problem Anchor by extracting immutable bottom-line problem statements, bottlenecks, non-goals, constraints, and success conditions.
- Phase 1: Build the Initial Proposal by (1) scanning grounding material for mechanism and gaps, (2) identifying the technical gap, (3) choosing the sharpest route (minimal vs frontier-native), (4) concretizing the method with detailed elements, (5) designing a minimal validation plan, and (6) writing the initial proposal to refine-logs/round-0-initial-proposal.md.
- Phase 2: External Method Review (Round 1) by sending the full proposal to GPT-5.4 for an elegance-first critique focused on method quality, with a detailed prompt that scores seven dimensions and provides actionable fixes. It requires saving the returned jobId, polling for completion, and storing the raw response in refine-logs/round-1-review.md.
- Phase 3: Parse Feedback and Revise the Method by extracting structured feedback from the review (e.g., Problem Fidelity, Method Specificity, etc.).
- Phase 4: (Not fully shown in the excerpt) likely includes implementing revisions and producing further refinement rounds until readiness.
When to use it
Use when the user wants to refine an approach, decompose a problem, or produce a concrete, implementable research method that remains simple, focused, and top-venue ready rather than vague or overbuilt.
What it can touch
- Tools and models are defined as:
- REVIEWER_MODEL =
claude-review(via local MCP bridge) for the external critique phase. - OUTPUT_DIR =
refine-logs/where all rounds and reports are stored.
- REVIEWER_MODEL =
- The workflow specifies file paths and naming conventions for all outputs (e.g., refine-logs/round-0-initial-proposal.md, refine-logs/round-1-review.md).
Caveats
- The method requires a problem anchor to be frozen and preserved across rounds.
- It relies on multi-round human or model reviews and expects the evaluator to provide concrete, actionable feedback.
- The process mandates saving raw reviewer responses and tracking job IDs for round-based status reporting.
> Override for Codex users who want **Claude Code**, not a second Codex agent, to act as the reviewer. Install this package **after** `skills/skills-codex/*`. # Research Refine: Problem-Anchored, Elegant, Frontier-Aware Plan Refinement Refine and concretize: **$ARGUMENTS** ## Overview Use this skill when the research problem is already visible but the technical route is still fuzzy. The goal is not to produce a bloated proposal or a benchmark shopping list. The goal is to turn a vague direction into a **problem -> focused method -> minimal validation** document that is concrete enough to implement, elegant enough to feel paper-worthy, and current enough to resonate in the foundation-model era. Four principles dominate this skill: 1. **Do not lose the original problem.** Freeze an immutable **Problem Anchor** and reuse it in every round. 2. **The smallest adequate mechanism wins.** Prefer the minimal intervention that directly fixes the bottleneck. 3. **One paper, one dominant contribution.** Prefer one sharp thesis plus at most one supporting contribution. 4. **Modern leverage is a prior, not a decoration.** When LLM / VLM / Diffusion / RL / distillation / inference-time scaling na
- Overview
- Constants
- Output Structure
- Workflow
- Phase 0: Freeze the Problem Anchor
- Phase 1: Build the Initial Proposal
- Phase 2: External Method Review (Round 1)
- Phase 3: Parse Feedback and Revise the Method
- Phase 4: Re-evaluation (Round 2+)
- Phase 5: Final Report and Logs
- Output Protocols
- Key Rules
- Composing with Other Skills
What does the research-refine skill do?
Turn a vague research direction into a problem-anchored, elegant, frontier-aware, implementation-oriented method plan via iterative GPT-5.4 review. Use when the user says \"refine my approach\", \"帮我细化方案\", \"decompose this problem\", \"打磨idea\", \"refine research plan\", \"细化研究方案\", or wants a concrete research method that stays simple, focused, and top-venue ready instead of a vague or overbuilt idea.
How do I install it?
Run `npx skills add majiayu000/claude-skill-registry --skill research-refine-wanshuiyin-auto-claude-code-res-2 --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From majiayu000/claude-skill-registry, a repository with 534 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.