Skill

The LLM Council

Turn one question into five expert opinions, then a single clear verdict

The Idea

Ask one AI a question and you get one answer. It might be brilliant. It might be average. The problem is you can't tell — because you only ever saw a single point of view.

The council fixes that. It runs your question through five independent advisors, each reasoning from a completely different angle. They peer-review each other's work. Then a chairman pulls it all together into one straight answer: where they agree, where they clash, and what you should actually do about it.

Install It (No Terminal Needed)

The easiest way: let Claude install it for you. Open a new chat — in Claude Code or on the web at claude.ai — and paste this in. Claude grabs the file, drops it in the right place, and tells you if it needs anything from you.

Paste this into Claude
Please install this Claude skill for me. The SKILL.md file lives in this GitHub repo: https://github.com/oliwoodman/llm-council-skill Set it up so I can start using it, and walk me through anything you need from me.

Once it's installed, trigger it in any conversation with a phrase like:

council thisrun the council on [your question]pressure-test thisstress-test thiswar room this

The Five Advisors

They're not personas — they're thinking styles, chosen because they create tension with each other. Contrarian versus Expansionist (downside versus upside). First Principles versus Executor (rethink it versus just do it). The Outsider sits in the middle keeping everyone honest.

1

The Contrarian

Hunts the fatal flaw

Actively looks for what's wrong, what's missing, what will fail. The friend who saves you from a bad deal by asking the questions you're avoiding.

2

The First Principles Thinker

Questions the question

Ignores the surface question and asks what you're actually trying to solve. Sometimes its most valuable output is "you're asking the wrong question entirely."

3

The Expansionist

Finds the upside

Looks for the bigger opportunity everyone else is missing. Doesn't care about risk — cares about what happens if this works even better than expected.

4

The Outsider

Has zero context

Reacts purely to what's in front of them. Catches the curse of knowledge — the things obvious to you but confusing to everyone else.

5

The Executor

Only cares what's doable

Looks at every idea through "OK, but what do you do Monday morning?" If it sounds brilliant but has no first step, the Executor says so.

How a Session Works

1

Frame the question

Claude scans your workspace for context — CLAUDE.md, any memory folder, past transcripts — then reframes your raw question into one neutral prompt all five advisors receive. No opinion added, just enough context to get specific answers instead of generic ones.

2

Convene the council

All five advisors run in parallel, each leaning fully into its angle. No hedging, no trying to be balanced — 150–300 words each, straight into the analysis.

3

Peer review

The responses are anonymised and shuffled, then each advisor reviews all five: which is strongest, which has the biggest blind spot, and what everyone missed. This is the step that makes it more than "ask five times."

4

The chairman's verdict

One agent synthesises everything into a single answer: where the council agrees, where it clashes, the blind spots peer review caught, a real recommendation, and the one thing to do first.

When to Use It

The council earns its keep when being wrong is expensive and the answer isn't obvious. If you already know the answer and just want a pat on the back — fair warning, it'll probably tell you things you don't want to hear. That's the point.

Good council questions

  • Should I launch a £97 workshop or a £497 course?
  • Which of these 3 positioning angles is strongest?
  • I'm thinking of pivoting from X to Y — am I crazy?
  • Here's my landing page copy. What's weak?
  • Should I hire a VA or build an automation first?

Not council questions

  • "What's the capital of France?" — one right answer, no perspectives needed.
  • "Write me a tweet" — a creation task, not a decision.
  • "Summarise this article" — processing, not judgment.

Bring it your next real decision.

Install it once, then the next time you're torn between two options, just say “council this” and let five advisors fight it out for you.

View the skill on GitHub →

Built by Ole Lehmann, based on Andrej Karpathy's LLM Council methodology. This is a lightly rewritten version, packaged to be easy to install and share.