MCP Consensus Server

A Model Context Protocol (MCP) server that implements an advisor-based consensus mechanism for collaborative problem-solving using multiple AI models.

Features

• Multi-Advisor Consensus System: Engages 5 specialized AI advisors in structured discussion
• Collaborative Problem-Solving: Advisors work together to find optimal solutions through debate and analysis
• Tool Integration: Advisors can request additional information through available tools
• Configurable Discussion Parameters: Adjustable rounds and consensus thresholds
• Real-time Discussion Logging: Colorful console output showing the consensus process
• Multiple AI Models: Utilizes different models (OpenAI, Anthropic, DeepSeek, Moonshot, Z-AI) for diverse perspectives

Installation

npm install @dakraid/mcp-consensus

Prerequisites

• Node.js 18+
• OpenRouter API key (set as OPENROUTER_API_KEY environment variable)

Usage

MCP Server Configuration

Add to your MCP client configuration:

{
  "mcpServers": {
    "consensus": {
      "command": "npx",
      "args": [
        "-y",
        "@dakraid/mcp-consensus"
      ],
      "env": {
        "OPENROUTER_API_KEY": "",
        "CONSENSUS_MAX_ROUNDS": "5",
        "CONSENSUS_THRESHOLD": "0.8"
      }
    }
  }
}

Available Tools

consensus

A multi-advisor consensus system that facilitates structured discussion and debate among general-purpose AI advisors to reach optimal solutions.

Parameters:

• problem (required): Detailed description of the problem to solve
• availableTools (required): Array of tool names available for research

Example Usage:

// Basic consensus
{
  "problem": "Should we adopt a remote-first work policy for our tech company?",
  "availableTools": ["web_search", "read_file"]
}

How It Works

1. Problem Presentation: The problem is presented to all 5 advisors simultaneously
2. Initial Analysis: Each advisor provides their analysis and proposed solution
3. Tool Requests: Advisors can request additional information through available tools
4. Multi-Round Discussion: Advisors engage in structured debate, considering each other's perspectives
5. Consensus Detection: The system monitors for agreement based on the configured threshold
6. Result Delivery: Returns the final consensus with complete discussion history

Advisors

The system includes 5 pre-configured advisors, each using different AI models:

• Advisor Alpha: Moonshot AI Kimi-k2
• Advisor Beta: DeepSeek Chat v3
• Advisor Gamma: Z-AI GLM-4.5
• Advisor Delta: OpenAI GPT-4.1
• Advisor Epsilon: Anthropic Claude Sonnet 4

Each advisor follows core principles of objectivity, collaboration, thoroughness, adaptability, and clarity.

Tool Request Format

Advisors can request additional information using this format:

TOOL_REQUEST: {"tool": "web_search", "parameters": {"query": "remote work productivity statistics"}, "reason": "I need current data on remote work effectiveness"}

Response Structure

The consensus tool returns:

{
  "status": "consensus_reached" | "max_rounds_reached" | "tool_requests_needed",
  "finalConsensus": "The agreed-upon solution",
  "totalRounds": 3,
  "discussionHistory": [...]
}

Development

# Clone the repository
git clone https://github.com/dakraid/mcp-consensus.git
cd mcp-consensus

# Install dependencies
npm install

# Build the project
npm run build

# Watch for changes
npm run watch

Configuration

Environment Variables

• OPENROUTER_API_KEY: Required API key for OpenRouter
• CONSENSUS_MAX_ROUNDS: Maximum number of discussion rounds (default: 5, range: 1-10)
• CONSENSUS_THRESHOLD: Agreement threshold for consensus detection (default: 0.8, range: 0.0-1.0)
• DISABLE_CONSENSUS_LOGGING: Set to "true" to disable console logging (default: false)

Customization

You can modify the advisor configurations in index.ts to:

• Change system prompts
• Use different AI models
• Add or remove advisors
• Adjust model parameters

License

MIT License - see LICENSE file for details.

Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests if applicable
5. Submit a pull request

Support

For issues and questions, please visit the GitHub Issues page.

Author

Created by @dakraid