Research Refine
Turns a vague research idea into a focused, implementable method plan via iterative reviewer feedback
- Stars
- 18,482
- Licence
- MIT
- Files
- 1 file
What it does
When installed, the agent takes a fuzzy research direction and works through a structured loop: it freezes the original problem as a fixed anchor, scans grounding papers, proposes the smallest adequate method, sends the draft to an external reviewer model (named GPT-6-Astra, reached through an MCP connection), and revises until the reviewer's score reaches 9 or five rounds pass. The output is a saved set of round files and a final proposal in a refine-logs folder, with checkpoint state so an interrupted session can resume.
Great for
- Researchers who have a problem in mind but an unclear technical approach
- Students drafting a paper method who want to avoid overbuilt ideas
- Teams wanting a documented, reviewed trail of how a research plan evolved
Skip it if
- Simple one-off brainstorming, since the multi-round review loop is slow and token-heavy
- Users without access to the external reviewer model or MCP servers, because the core loop depends on them and the skill has no fallback described in the file
- Tasks needing guaranteed-accurate literature claims, because it relies on web search and model judgement that the file does not verify
Try asking
- “Refine my approach for detecting drift in sensor data before I write the methods section”
- “Decompose this problem: reducing hallucinations in a retrieval-augmented QA system”
- “帮我细化方案 for a low-cost fine-tuning method for small clinical language models”
- “Turn this vague idea into a concrete research plan that stays simple and focused”
Install in 3 steps
- 1
Download the skill
The ZIP contains the 1 original files, the MIT licence and an attribution note.
research-refine.zip - 2
Add it to your agent
Claude Code (all projects)
mkdir -p ~/.claude/skills && unzip research-refine.zip -d ~/.claude/skills/Claude Code / Agent SDK (this project only)
mkdir -p .claude/skills && unzip research-refine.zip -d .claude/skills/Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.
- 3
Ask for the task
No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.
Get better results
- Provide the problem and a rough approach in a clear 'problem | approach' form so the Problem Anchor is accurate
- Check the refine-logs folder after each run, since the score history and final report show the reasoning behind changes
- Keep the reviewer backend configured in .aris/reviewer.json to avoid falling back to defaults you did not choose
What's inside
- SKILL.md
Redistributed unmodified from wanshuiyin/Auto-claude-code-research-in-sleep under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.