LLM Council

Run any question, idea, or decision through a council of 5 AI advisors who independently analyze it, peer-review each other anonymously, and synthesize a final verdict.

Based on Andrej Karpathy's LLM Council methodology.

How it Works

1. Frame — Your question gets enriched with workspace context
2. Convene — 5 advisors analyze independently in parallel (Contrarian, First Principles, Expansionist, Outsider, Executor)
3. Peer Review — Advisors review each other's work anonymously
4. Synthesize — A chairman produces the final verdict with agreements, clashes, blind spots, and a clear recommendation
5. Report — Visual HTML report + full markdown transcript

The Five Advisors

Advisor	Thinking Style
The Contrarian	Finds what's wrong, missing, or will fail
The First Principles Thinker	Strips assumptions, rebuilds from ground up
The Expansionist	Finds upside everyone else is missing
The Outsider	Zero context, fresh eyes, catches curse of knowledge
The Executor	What do you do Monday morning?

Triggers

• council this
• run the council
• war room this
• pressure-test this
• stress-test this
• debate this

Install

Copy SKILL.md to your Claude Code skills folder:

# Project-level
cp SKILL.md .claude/skills/llm-council/SKILL.md

# Global (all projects)
cp SKILL.md ~/.claude/skills/llm-council/SKILL.md

Or for OpenClaw users, place in your workspace skills/ directory.

Output

Every council session produces:

• council-report-[timestamp].html — Visual report for scanning
• council-transcript-[timestamp].md — Full transcript for reference

When to Use

✅ Decisions with real stakes and trade-offs ✅ Multiple viable options where being wrong is expensive ✅ Ideas that need pressure-testing from different angles

❌ Factual lookups with one right answer ❌ Creation tasks (write me a tweet) ❌ Processing tasks (summarize this)

Credits

• Methodology: Andrej Karpathy
• Skill adaptation: Bruno Okamoto
• Source article: Google Doc

License

MIT