MCP Router

Offers automatic model routing and decision support for Cursor MCP, with 17 models and strategy-based routing.

python

1

GitHub Stars

python

Language

5 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "ai-castle-labs-mcp-router": {
      "command": "python3",
      "args": [
        "/path/to/mcp-router/src/mcp_server.py"
      ],
      "env": {
        "ENV": "{}"
      }
    }
  }
}

You use the MCP Router to automatically select the best machine learning model for each task inside the Cursor IDE. It analyzes your query, weighs strategy and cost, and routes execution to the most suitable model, all without you managing keys or separate API calls.

How to use

You interact with the MCP Router through the Cursor agent tools. When you have a task, describe it clearly and ask for a model recommendation. You can then execute or compare suggested models, or request a list of all available models. Use strategy presets to prioritize quality, speed, cost, or a balanced approach. You can also inspect routing statistics to understand past decisions.

How to install

Prerequisites: ensure Python is installed on your system.

Step 1: Clone the MCP Router repository.

Step 2: Change into the project directory.

Step 3: Install dependencies.

git clone https://github.com/AI-Castle-Labs/mcp-router.git
cd mcp-router
pip install -r requirements.txt
pip install mcp  # MCP SDK for Cursor integration

Step 4: Configure Cursor to use the MCP Router as a local, stdio-based server.

{
  "version": "1.0",
  "mcpServers": {
    "mcp_router": {
      "command": "python3",
      "args": ["/path/to/mcp-router/src/mcp_server.py"],
      "env": {}
    }
  }
}

Step 5: Place the Cursor configuration in your home directory to point Cursor at the MCP Router. You will add the above mcp_router server entry to your Cursor configuration file.

Step 6: Restart Cursor to load the new MCP Router integration. You will see the MCP Router appear as an agent tool in your Cursor environment.

Additional notes

Only the local stdio configuration is provided here since there is no remote MCP URL shown. The MCP Router runs as a local Python process and does not require external API keys. All model routing decisions occur inside the Router and Cursor handles the integration.

Available tools

route_query

Route a query and get model recommendation without executing the task.

get_model_recommendation

Get a single model recommendation for a given task description.

list_models

List all registered MCP models available for routing.

get_routing_stats

Retrieve usage statistics and historical routing decisions.

analyze_query

Analyze query characteristics such as task type, complexity, and requirements.