Think
Structured agentic development, shipped as Agent Skills - plan, triage, build, and review with autonomous sub-agents, human approval gates, and automatic review on every diff. Runs in Claude Code, Cursor, Codex, Gemini, Mistral, pi, and any Agent Skills harness.
npx -y skills add markstent/muster --skill thinkAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
(Muster) Talk through your idea, one question at a time, until every open question is settled - so the plan is solid before any code gets written. Use when you say "I want to build X", "I have an idea", or want to pressure-test a plan. Writes nothing. Ends by pointing you to /spec.
SKILL.md
1.8 KB, 348 tokens by cl100k_base, as published. Nobody here has run it
Think
Ask me about every part of this plan, one question at a time, until we both understand it the same way. Work through every open question - and the new ones each answer raises - settling them one at a time.
For each question, provide your recommended answer. Ask one question at a time.
If a question can be answered by exploring the codebase, explore the codebase instead of asking me.
What to resolve
Don't stop until every one of these is settled. Chase any new questions that open up along the way.
- The exact problem, from my perspective. Why this, why now?
- Who the user is. For personal projects: which version of me, doing what?
- The single most important behaviour.
- What "done" looks like - how I would verify it worked.
- What I am explicitly NOT building.
- The riskiest assumption in the plan.
- What existing code or systems this touches.
Closing
When every question is settled, recap it back to me in plain terms: the problem, what we'll build (what, not how), how we'll know it works, and what we're not building. Ask whether that matches what I meant.
If I correct something: settle the new question, then recap again. If I confirm, end with:
We're agreed on what to build.
**Next:** run /spec
Rules
- Treat vague answers ("roughly", "something like that") as unresolved. Ask again.
- Do not propose implementations unless I ask.
- Write nothing permanent. The point of this session is that we agree on what to build, not a document.
What ships with it
Read from the repository
Just SKILL.md. No reference files, no scripts.