AI Collaboration MCP Server

Claude Code AI Collaboration MCP Server

A powerful Model Context Protocol (MCP) server that enables AI collaboration through multiple providers with advanced strategies and comprehensive tooling.

🌟 Features

🤖 Multi-Provider AI Integration

• DeepSeek: Primary provider with optimized performance
• OpenAI: GPT models integration
• Anthropic: Claude models support
• O3: Next-generation model support

🚀 Advanced Collaboration Strategies

• Parallel: Execute requests across multiple providers simultaneously
• Sequential: Chain provider responses for iterative improvement
• Consensus: Build agreement through multiple provider opinions
• Iterative: Refine responses through multiple rounds

🛠️ Comprehensive MCP Tools

• collaborate: Multi-provider collaboration with strategy selection
• review: Content analysis and quality assessment
• compare: Side-by-side comparison of multiple items
• refine: Iterative content improvement

📊 Enterprise Features

• Caching: Memory and Redis-compatible caching system
• Metrics: OpenTelemetry-compatible performance monitoring
• Search: Full-text search with inverted indexing
• Synthesis: Intelligent response aggregation

🚀 Quick Start

📖 New to MCP? Check out our Quick Start Guide for a 5-minute setup!

Prerequisites

• Node.js 18.0.0 or higher
• pnpm 8.0.0 or higher
• TypeScript 5.3.0 or higher

Installation

# Clone the repository
git clone https://github.com/atsuki-sakai/ai_collaboration_mcp_server.git
cd ai_collaboration_mcp_server

# Install dependencies
pnpm install

# Build the project
pnpm run build

# Run tests
pnpm test

Configuration

1. Environment Variables:

   # Required: Set your API keys
   export DEEPSEEK_API_KEY="your-deepseek-api-key"
   export OPENAI_API_KEY="your-openai-api-key"
   export ANTHROPIC_API_KEY="your-anthropic-api-key"
   
   # Optional: Configure other settings
   export MCP_DEFAULT_PROVIDER="deepseek"
   export MCP_PROTOCOL="stdio"
   

2. Configuration Files:

   • config/default.yaml: Default configuration
   • config/development.yaml: Development settings
   • config/production.yaml: Production settings

Running the Server

# Start with default settings
pnpm start

# Start with specific protocol
node dist/index.js --protocol stdio

# Start with custom providers
node dist/index.js --providers deepseek,openai --default-provider deepseek

# Enable debug mode
NODE_ENV=development LOG_LEVEL=debug pnpm start

🔗 Claude Code Integration

Connecting to Claude Code

To use this MCP server with Claude Code, you need to configure Claude Code to recognize and connect to your server.

1. Automated Setup (Recommended)

Use the automated setup script for easy configuration:

# Navigate to your project directory
cd /Users/atsukisakai/Desktop/ai_collaboration_mcp_server

# Run automated setup with your DeepSeek API key
./scripts/setup-claude-code.sh --api-key "your-deepseek-api-key"

# Or with multiple providers
./scripts/setup-claude-code.sh \
  --api-key "your-deepseek-key" \
  --openai-key "your-openai-key" \
  --anthropic-key "your-anthropic-key"

# Alternative using pnpm
pnpm run setup:claude-code -- --api-key "your-deepseek-key"

The setup script will:

• ✅ Build the MCP server
• ✅ Create Claude Code configuration file
• ✅ Test the server connection
• ✅ Provide next steps

1b. Manual Setup

If you prefer manual setup:

# Navigate to your project directory
cd /Users/atsukisakai/Desktop/ai_collaboration_mcp_server

# Install dependencies and build
pnpm install
pnpm run build

# Set your DeepSeek API key
export DEEPSEEK_API_KEY="your-deepseek-api-key"

# Test the server
pnpm run verify-deepseek

2. Configure Claude Code

Create or update the Claude Code configuration file:

Note: There are two server options:

• simple-server.js - Simple implementation with DeepSeek only (recommended for testing)
• index.js - Full implementation with all providers and features

macOS/Linux:

# Create config directory if it doesn't exist
mkdir -p ~/.config/claude-code

# Create configuration file (simple server - recommended for testing)
cat > ~/.config/claude-code/claude_desktop_config.json << 'EOF'
{
  "mcpServers": {
    "ai-collaboration": {
      "command": "node",
      "args": ["/Users/atsukisakai/Desktop/ai_collaboration_mcp_server/dist/simple-server.js"],