You're in · the skill is yours
LLM Council.
AI is trained to agree with you. Ask it about your idea and it says “great idea!” even when it isn't. This free Claude Code skill fixes that: it convenes five sub-agents, each built to challenge your thinking from a different angle, then hands you one honest verdict, including the case against.
Build with me
Get the skills, prompts, and agent systems behind everything I ship.
Get it running in 30 seconds
Install the skill
git clone https://github.com/jsath/llm-council.git
cp -r llm-council/skills/llm-council ~/.claude/skills/Then in Claude Code just type /llm-council <your idea> or run this past the council: …
The five members
The Contrarian
Argues the strongest case against your idea. No softening.
The Industry Outsider
Asks the naive questions experts are too close to see.
The Practitioner
What real experience says actually breaks or costs more.
The Risk Officer
Failure modes, downside, and the worst realistic outcome.
The Customer
Would anyone actually want it, pay for it, or switch?
Inspired by Andrej Karpathy's open-source llm-council idea: don't ask one model, ask several with different roles and let them argue.
Want the systems, not just the skill?
I build stuff like this every week.
Inside AI Operators I share the exact skills, prompts, and agent systems behind everything I ship, plus weekly calls where I'll help you set them up. Come build with us.