council

Consult a multi LLM council for deliberation, debate, voting, critique, or verification. Uses GPT 5.4, Gemini 2.5, and Claude as peers.

16 stars

Best use case

council is best used when you need a repeatable AI agent workflow instead of a one-off prompt.

Consult a multi LLM council for deliberation, debate, voting, critique, or verification. Uses GPT 5.4, Gemini 2.5, and Claude as peers.

Teams using council should expect a more consistent output, faster repeated execution, less prompt rewriting.

When to use this skill

  • You want a reusable workflow that can be run more than once with consistent structure.

When not to use this skill

  • You only need a quick one-off answer and do not need a reusable workflow.
  • You cannot install or maintain the underlying files, dependencies, or repository context.

How council Compares

Feature / AgentcouncilStandard Approach
Platform SupportNot specifiedLimited / Varies
Context Awareness High Baseline
Installation ComplexityUnknownN/A

Frequently Asked Questions

What does this skill do?

Consult a multi LLM council for deliberation, debate, voting, critique, or verification. Uses GPT 5.4, Gemini 2.5, and Claude as peers.

Where can I find the source code?

You can find the source code on GitHub using the link provided at the top of the page.

SKILL.md Source

# LLM Council Skill

You have access to the LLM Council MCP tools. Use them to orchestrate multimodel deliberation.

## Available Tools

1. **council_deliberate**. Full council deliberation with configurable protocol
2. **council_vote**. Quick voting: all models answer, then anonymously rank each other
3. **council_debate**. Structured debate with adaptive stopping
4. **council_critique**. Peer critique or adversarial redteaming
5. **council_verify**. Multi agent verification of an answer
6. **council_estimate_cost**. Estimate cost before running
7. **council_status**. Check which providers are available
8. **council_configure**. Update default council composition

## Usage Patterns

### Quick consensus
Use `council_vote` for straightforward questions where you want the best answer selected by peer review.

### Deep analysis
Use `council_debate` with `adaptiveStop: true` for complex questions that benefit from iterative refinement. Set `maxRounds: 3` for thorough analysis.

### Verification
Use `council_verify` to check if an answer is correct by having multiple models independently verify it.

### Stress testing
Use `council_critique` with `redTeam: true` to find flaws, edge cases, and failure modes in responses.

### Full deliberation
Use `council_deliberate` with `protocol: "synthesize"` for the chairman synthesis pattern. All models answer, then a chairman produces an authoritative synthesis.

## Best Practices (from research)
* Scale agents (more models), not rounds (more debate turns)
* Heterogeneous models (mix providers) outperform homogeneous ones
* Anonymous peer review prevents model favoritism bias
* Always check cost estimate before large council runs
* Default council (GPT 5.4 + Gemini 2.5 Pro + Claude Sonnet 4.6) provides optimal diversity

## Example Invocations

User: "What are the tradeoffs of microservices vs monolith?"
→ Use council_vote for quick consensus

User: "Is this algorithm correct? [code]"
→ Use council_verify

User: "Debate whether Rust or Go is better for this use case"
→ Use council_debate with maxRounds: 2

User: "Red team this API design"
→ Use council_critique with redTeam: true

Related Skills

We are still matching the closest adjacent skills for this page. In the meantime, continue through the full directory.