question-refiner — Claude Code skill: what it does, how to install
question-refiner transforms a vague research question into a structured deep-research brief. Asks clarifying questions, collects requirements for format, sources and scope, then generates a ready-to-use prompt.
no license ★ 475 updated 2026-10-13
npx skills add cnfjlhj/ai-collab-playbook --skill question-refiner -a claude-codeWhat question-refiner does
The skill operates in two stages. First, it asks a series of questions: what exactly you are researching, what output format you need, whether there are geographic or time constraints, which sources count as credible. Once all details are collected, it assembles a structured prompt with TASK, CONTEXT, SPECIFIC QUESTIONS, KEYWORDS, CONSTRAINTS, OUTPUT FORMAT and FINAL INSTRUCTIONS sections. A seven-point checklist verifies that every field is filled concretely.
The skill has a hard rule: do not answer the research question itself, only refine and formulate. If you provide incomplete answers, it asks the next round of questions. It will not generate the prompt without answers to clarifying questions: the author wanted to replace ChatGPT o3 for this task, and the strict sequence prevents delivering a vague brief.
The skill does not run the actual research, search literature or write reports.
What it helps with
- turn a vague idea into a concrete research specification
- gather all requirements for format, sources and scope before starting work
- avoid the need to go to ChatGPT for question refinement
- verify that the prompt includes timeframe, geography and source types
Example prompts for the agent
- I need to research the AI tooling market for developers, but I don't know how to phrase it properly
- Help me build a prompt for research on remote work's impact on team productivity
- Ask me questions so I can figure out what I actually need to know about blockchain in logistics
How to install question-refiner
1Skills CLI (Vercel)
npx skills add cnfjlhj/ai-collab-playbook --skill question-refiner -a claude-code2Manually: put the files in place
~/.claude/skills/question-refiner/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/question-refiner/
1Skills CLI (Vercel)
npx skills add cnfjlhj/ai-collab-playbook --skill question-refiner -a codex2Manually: put the files in place
.agents/skills/question-refiner/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add cnfjlhj/ai-collab-playbook --skill question-refiner -a cursor2Manually: put the files in place
.cursor/skills/question-refiner/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install cnfjlhj/ai-collab-playbook skills/full/question-refiner2Manually: put the files in place
.github/skills/question-refiner/SKILL.md1Skills CLI (Vercel)
npx skills add cnfjlhj/ai-collab-playbook --skill question-refiner -a gemini-cli2Platform CLI
gemini skills install https://github.com/cnfjlhj/ai-collab-playbook --path skills/full/question-refinerManually: put the files in place
.devin/skills/question-refiner/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about question-refiner
How to install question-refiner in Claude Code?
Run npx skills add cnfjlhj/ai-collab-playbook --skill question-refiner -a claude-code, then open ~/.claude/skills/question-refiner/SKILL.md.
What does question-refiner do?
Transforms a vague research question into a structured prompt with sections for task, context, keywords, constraints and output format. First asks clarifying questions, then assembles a ready specification.
When should I skip question-refiner?
If you already have a paper draft or a ready research specification. The skill is useful at the question-formulation stage, not for executing the research itself.