AI Counsel is an innovative multi-consensus protocol (MCP) server designed to facilitate constructive debates among artificial intelligence models, enabling them to refine their positions through multiple rounds of deliberation. This robust server is built to promote true deliberative consensus, where models engage in actual debate, view each other's responses, and evolve their arguments across different rounds.

Key Features

• Dual Modes: Choose between quick (single-round discussions) or conference (multi-round debates) to suit specific needs.
• Mixed Adapters: Integrate various AI models through CLI tools (Claude, Codex, etc.) and HTTP services (Ollama, LM Studio).
• Auto-Convergence: The system automatically stops deliberation once opinions stabilize, optimizing both time efficiency and API costs.
• Structured Voting: Models can cast votes complete with confidence levels and rationales to determine consensus outcomes.
• Evidence-Based Deliberation: Ground discussions in reality by allowing models to read files, search code, and execute commands during deliberations.
• Local Model Support: Facilitate private discussions and maintain data security with zero API costs by utilizing local models like Ollama and others.
• Decision Graph Memory: AI Counsel learns from past deliberations, providing context that helps models converge faster in future discussions.

Example Usage

To demonstrate, here's a sample code snippet for a deliberation:

mcp__ai-counsel__deliberate({  
  question: "Should we use REST or GraphQL for our new API?",  
  participants: [  
    {cli: "claude", model: "claude-sonnet-4-5-20250929"},  
    {cli: "codex", model: "gpt-5-codex"},  
    {cli: "gemini", model: "gemini-2.5-pro"}  
  ],  
  mode: "conference",  
  rounds: 3  
})  

The results from this deliberation can be viewed in a full transcript, offering insights into the consensus achieved by the participating models.

What Makes AI Counsel Different

AI Counsel stands out by enabling models to engage in genuine debates rather than simply gathering parallel opinions. This results in:

• A clear audit trail with AI-generated summaries for transparency.
• Automatic halting of discussions when consensus is reached, optimizing costs.
• A system that can handle evidence-based decision-making, making it suitable for technical and strategic evaluations.

With a focus on ensuring privacy and accommodating both cloud and local model setups, AI Counsel provides a comprehensive solution for collaborative AI decision-making.