A complete Model Context Protocol (MCP) server implementation for CRM operations with advanced AI features

This is a real MCP server that implements the Model Context Protocol specification. It provides AI-powered CRM functionality through standardized MCP tools that can be used by AI agents, Cursor, and other MCP-compatible clients.

• ✅ Real MCP Server - Follows Model Context Protocol specification
• 🧠 AI-Powered Analytics - Revenue prediction, contact scoring, relationship mapping
• 🔌 MCP Tool Integration - 10 comprehensive tools for CRM operations
• 📊 Advanced Insights - Company analysis, trend detection, smart recommendations
• 💬 AI Chat Interface - Natural language queries for CRM data
• 🕸️ Relationship Mapping - Visual network of contact connections

┌─────────────────────────────────────────────────────────────┐
│                    MCP Server Layer                        │
├─────────────────────────────────────────────────────────────┤
│  • ListToolsRequestSchema  • CallToolRequestSchema         │
│  • StdioServerTransport    • Server Instance               │
└─────────────────────────────────────────────────────────────┘
                              │
┌─────────────────────────────────────────────────────────────┐
│                  AI Analytics Engine                        │
├─────────────────────────────────────────────────────────────┤
│  • Revenue Prediction      • Contact Scoring               │
│  • Relationship Mapping    • Smart Recommendations         │
└─────────────────────────────────────────────────────────────┘
                              │
┌─────────────────────────────────────────────────────────────┐
│                    Supabase Backend                        │
├─────────────────────────────────────────────────────────────┤
│  • PostgreSQL Database     • Real-time Subscriptions       │
│  • Row Level Security     • API Authentication            │
└─────────────────────────────────────────────────────────────┘

• Node.js 18+
• Supabase account and project
• MCP-compatible client (Cursor, Claude Desktop, etc.)

npm install

Create .env file:

# Supabase Configuration
SUPABASE_URL=https://your-project.supabase.co
SUPABASE_KEY=your-supabase-anon-key

# Optional: Server Configuration
PORT=3000

Create the contacts table in Supabase:

CREATE TABLE contacts (
  id UUID DEFAULT gen_random_uuid() PRIMARY KEY,
  name TEXT NOT NULL,
  email TEXT NOT NULL,
  company TEXT,
  revenue INTEGER DEFAULT 0,
  created_at TIMESTAMP WITH TIME ZONE DEFAULT NOW()
);

-- Enable Row Level Security
ALTER TABLE contacts ENABLE ROW LEVEL SECURITY;

-- Create policy for public access (adjust as needed)
CREATE POLICY "Allow public access" ON contacts FOR ALL USING (true);

1. get_contacts - Retrieve contacts with AI scoring
2. add_contact - Add new contact with AI insights
3. get_top_clients - Get top clients by revenue
4. search_contacts - Search contacts by name/email

5. get_ai_insights - Comprehensive AI analytics
6. predict_revenue - AI revenue prediction for contacts
7. get_relationship_map - Network visualization data
8. get_analytics - Detailed performance metrics
9. get_recommendations - AI-powered action suggestions

10. ai_chat - Chat with AI assistant for CRM queries

Add to your Cursor MCP configuration:

{
  "mcpServers": {
    "crm-assistant": {
      "command": "node",
      "args": ["/path/to/crm-mcp/server.js"],
      "env": {
        "SUPABASE_URL": "your-supabase-url",
        "SUPABASE_KEY": "your-supabase-key"
      }
    }
  }
}

// MCP Tool Call
{
  "name": "get_contacts",
  "arguments": {
    "limit": 10
  }
}

// MCP Tool Call
{
  "name": "add_contact",
  "arguments": {
    "name": "John Doe",
    "email": "[email protected]",
    "company": "TechCorp",
    "revenue": 15000
  }
}

// MCP Tool Call
{
  "name": "ai_chat",
  "arguments": {
    "message": "Who are my top 5 tech clients?"
  }
}

// Factors considered:
- Company size (length > 10 chars = 1.5x multiplier)
- Email domain (business vs personal)
- Name length (longer names = higher potential)
- Current revenue (base for prediction)

// Formula:
basePrediction = currentRevenue * 1.2
multiplier = companySize * emailDomain * nameLength
predictedRevenue = basePrediction * multiplier

// Scoring breakdown:
- Revenue Factor: 40% weight (max 40 points)
- Company Factor: 30% weight (tech companies get bonus)
- Email Domain: 20% weight (business domains preferred)
- Name Length: 10% weight (longer names = higher score)

// Connection types detected:
- Colleague: Same company (strength: 0.9)
- Domain: Same email domain (strength: 0.7)
- High-value: Similar revenue ranges (strength: 0.6)
- Tech-network: Both in tech companies (strength: 0.5)
- Strategic: Random connections (strength: 0.3)

The AI chat interface understands natural language queries:

• Count Queries: "How many contacts do I have?"
• Revenue Analysis: "What's my total revenue?"
• Top Clients: "Who is my most profitable client?"
• Tech Filtering: "Show me my top 10 tech clients"
• Analytics: "Give me AI insights on my contacts"
• Recommendations: "What should I focus on?"

User: "Who is my most profitable person?"
AI: "💰 Your most profitable person is John Smith with $25,000 in revenue!"

User: "Show me my top 5 tech clients"
AI: "🚀 Your top 5 tech clients:
1. Jane Doe - $15,000
2. Bob Wilson - $12,000
3. Alice Brown - $8,500
..."

User: "How many contacts do I have?"
AI: "📊 You have 47 contacts in your CRM!"

const server = new Server(
  {
    name: 'crm-assistant-mcp',
    version: '1.0.0',
  },
  {
    capabilities: {
      tools: {},
    },
  }
);

{
  name: 'get_contacts',
  description: 'Get all contacts from the CRM with AI scoring and insights',
  inputSchema: {
    type: 'object',
    properties: {
      limit: {
        type: 'number',
        description: 'Maximum number of contacts to return (default: 50)',
        default: 50
      }
    }
  }
}

server.setRequestHandler(CallToolRequestSchema, async (request) => {
  const { name, arguments: args } = request.params;
  // Tool implementation...
});

npm run dev

npm start

🚀 MCP CRM Assistant Server Started! 🚀
🧠 AI Features: Analytics, Predictions, Relationship Mapping
⚡ Real-time: Live updates, notifications, smart insights
🎯 Ready for INSANE CRM operations via MCP!

1. "Please set SUPABASE_URL and SUPABASE_KEY"

   • Ensure .env file exists with correct values
   • Check Supabase project is active

2. "Unknown tool" errors

   • Verify MCP client is properly configured
   • Check tool names match exactly

3. Database connection errors

   • Verify Supabase URL and key are correct
   • Ensure contacts table exists
   • Check RLS policies allow access

DEBUG=mcp* node server.js

• Contacts: 50 per request (configurable)
• Search Results: 50 per query
• AI Processing: Real-time (no caching)

• Implement caching for AI insights
• Add pagination for large datasets
• Optimize relationship mapping algorithm
• Add database indexing

• Real-time Updates: WebSocket integration for live data
• Advanced AI: Machine learning models for predictions
• Integration APIs: Connect with external CRM systems
• Custom Scoring: User-defined scoring algorithms
• Bulk Operations: Batch contact operations
• Export/Import: Data migration tools

• Model Context Protocol Specification
• MCP SDK Documentation

• Supabase Documentation
• PostgreSQL Documentation

• Revenue Prediction Algorithms
• Customer Scoring Models

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

MIT License - see LICENSE file for details

This MCP server transforms your CRM into an AI-powered intelligence platform. By implementing the Model Context Protocol, it seamlessly integrates with AI agents and development tools, providing powerful CRM operations through a standardized interface.

Ready to revolutionize your CRM with AI? Let's get started! 🚀