DynaRoute MCP Server

A Model Context Protocol (MCP) server that provides intelligent chat completions with automatic model routing and cost optimization through DynaRoute.

Features

• 🧠 Intelligent Model Routing: Automatically routes queries to the most cost-effective model while maintaining quality
• 💰 Cost Optimization: Save up to 70% compared to premium models like GPT-4o
• 📊 Detailed Analytics: Get comprehensive metrics including token usage, costs, and performance data
• 🔍 Routing Intelligence: See exactly which model was selected and why
• ⚡ Easy Integration: Works seamlessly with Claude Desktop, Cursor, and other MCP-compatible clients

Installation

Install via pip:

pip install dynaroute-mcp-server

Quick Start

1. Get your DynaRoute API key from DynaRoute

2. Configure Claude Desktop by adding this to your claude_desktop_config.json:

   {
     "mcpServers": {
       "dynaroute": {
         "command": "uv",
         "args": ["tool", "run", "dynaroute-mcp-server"],
         "env": {
           "DYNAROUTE_API_KEY": "YOUR_API_KEY"
         }
       }
     }
   }
   

3. Restart Claude Desktop and start using DynaRoute!

Usage

Once configured, you can use DynaRoute in Claude Desktop by:

• Asking for AI responses: "Use DynaRoute to explain quantum computing"
• Requesting cost analysis: "Get a cost-optimized response about machine learning"
• Explicit tool usage: "Use the DynaRoute tool to answer this question"

Example Output

When you use the DynaRoute tool, you'll get:

🤖 DynaRoute Response:
[Your AI response content here]

📊 PERFORMANCE METRICS:
• Model Used: gcp-gemini-2.0-flash-thinking
• Response Time: 2.5 seconds
• Request ID: chatcmpl-xyz123

🔢 TOKEN USAGE:
• Input Tokens: 25
• Output Tokens: 150
• Total Tokens: 175

💰 COST BREAKDOWN:
• Input Cost: $0.000012
• Output Cost: $0.000045
• Total Cost: $0.000057

📈 COST SAVINGS vs GPT-4o:
• GPT-4o Equivalent Cost: $0.000175
• Your Actual Cost: $0.000057
• Money Saved: $0.000118
• Savings Percentage: 67.4%

✅ EFFICIENCY SUMMARY: DynaRoute saved you 67.4% compared to GPT-4o while maintaining quality!

Configuration

Environment Variables

• DYNAROUTE_API_KEY: Your DynaRoute API key (required)

Tool Parameters

The dynaroute_chat tool accepts:

• messages: Array of conversation messages (required)
• level_override: Override complexity level 1-5 (optional)

Claude Desktop Configuration

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS):

{
  "mcpServers": {
    "dynaroute": {
      "command": "uv",
      "args": ["tool", "run", "dynaroute-mcp-server"],
      "env": {
        "DYNAROUTE_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

Other MCP Clients

Cursor IDE

Add to your Cursor settings:

{
  "mcp": {
    "servers": {
      "dynaroute": {
        "command": "uv",
        "args": ["tool", "run", "dynaroute-mcp-server"],
        "env": {
          "DYNAROUTE_API_KEY": "YOUR_API_KEY"
        }
      }
    }
  }
}

Custom Usage

You can also import and use the server programmatically:

import asyncio
from dynaroute_mcp import DynaRouteMCPServer

async def main():
    server = DynaRouteMCPServer(api_key="your_api_key")
    await server.run()

asyncio.run(main())

Requirements

• Python 3.8+
• DynaRoute API key
• MCP-compatible client (Claude Desktop, Cursor, etc.)

Dependencies

• mcp: Model Context Protocol implementation
• dynaroute-client: Official DynaRoute Python client

License

MIT License

Support

• 📧 Email: abraar237@gmail.com
• 🐛 Issues: GitHub Issues
• 📖 Documentation: GitHub README

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

Changelog

1.0.0

• Initial release
• DynaRoute integration with MCP
• Cost optimization and analytics
• Claude Desktop support