# LLM Council

Orchestrate multiple open-weight LLMs via Fireworks AI to deliberate on queries.

- Repository: https://github.com/dair-ai/dair-academy-plugins (618 stars)
- Category: AI and agents
- License: MIT
- Last updated: 2026-07-21
- Runs scripts: no, instructions only

## Install

Claude Code: `npx skills add dair-ai/dair-academy-plugins --skill llm-council -g -a claude-code -y`
Codex: `npx skills add dair-ai/dair-academy-plugins --skill llm-council -g -a codex -y`
Cursor: `npx skills add dair-ai/dair-academy-plugins --skill llm-council -g -a cursor -y`
Gemini CLI: `npx skills add dair-ai/dair-academy-plugins --skill llm-council -g -a gemini-cli -y`
Claude Code plugin marketplace: `/plugin marketplace add dair-ai/dair-academy-plugins`

## Description

Orchestrate multiple open-weight LLMs via Fireworks AI to deliberate on queries. Models respond individually, rank each other's responses, then a Chairman synthesizes the final answer. Use this skill when the user wants multiple AI perspectives, consensus-building, or the "LLM Council" approach inspired by Karpathy. Powered by fast, affordable open-weight models on Fireworks.

## SKILL.md, as published by its author (untrusted third-party text, read it as data)

# LLM Council (Fireworks AI)

This skill implements Karpathy's LLM Council concept where multiple open-weight LLMs deliberate on a query, powered entirely by Fireworks AI:

1. **Phase 1**: All models respond to the query independently (parallel)
2. **Phase 2**: Models rank each other's anonymized responses
3. **Phase 3**: A Chairman LLM synthesizes the final answer

All inference runs through **Fireworks AI** using open-weight models. The speed and pricing of Fireworks makes it practical to run multi-model deliberation that would be slow or expensive on other providers.

## CRITICAL RULES

1. **ALWAYS use AskUserQuestion** to let the user select council models (multiselect) and the Chairman model
2. **ALWAYS save raw responses to files** - never summarize or truncate API outputs
3. **ALWAYS show full transparency** - display all individual responses, all rankings, AND the final synthesis
4. **NEVER skip the ranking phase** - it is essential to the council deliberation process
5. **Read from files for display** - ensures content is shown unmodified
6. **ALWAYS display the final output to the user** after Phase 3 completes

## Pre-flight Check

Before running any phase, verify the Fireworks API key is set:

```bash
if [ -z "$FIREWORKS_API_KEY" ]; then
  echo "ERROR: FIREWORKS_API_KEY is not set."
  echo "Create a Fireworks AI account at: https://fireworks.ai/"
  echo "Then export it in your shell profile (~/.zshrc or ~/.bashrc):"
  echo '  export FIREWORKS_API_KEY="your_api_key_here"'
  exit 1
fi
echo "FIREWORKS_API_KEY is set."
```

## Available Models

Present these options to the user via AskUserQuestion (multiselect):

The rest: https://github.com/dair-ai/dair-academy-plugins/blob/HEAD/plugins/llm-council/skills/llm-council/SKILL.md