Agent Council
Gather opinions from multiple AI Agents and synthesize responses.
Inspired by Karpathy's LLM Council (https://github.com/karpathy/llm-council)
Execution Instructions
When this skill is invoked, follow these steps:
Step 1: Determine Query Type
Parse the user request:
- •Pattern "council AGENT_NAME: MESSAGE" means single agent query
- •Otherwise means full council query
Step 2: Execute Query
For SINGLE AGENT query (council gemini: ..., council claude: ...):
- •Use Bash tool to pipe the message to the specified agent CLI
- •Gemini (gemini) agent: pipe message to "
gemini --model pro 2>/dev/null" command. - •Claude (gemini) agent: pipe message to "
claude -p" command.
For FULL COUNCIL query (summon the council, ask other AIs):
- •Use Bash tool to run the
council.shscript located at$HOME/.claude/skills/agent-council/scripts/council.sh - •Pass the question as an argument
Step 3: Present Results
- •Prefix Gemini responses with diamond emoji
- •Prefix Claude responses with brain emoji
- •If multiple responses, synthesize as Chairman
Agent Commands Reference
See council.config.yaml for configured agents and their CLI commands.
Requirements
Each agent CLI must be installed and authenticated.