OpenRouter MCP Client for Cursor

An MCP client for Cursor that uses OpenRouter.ai to access multiple AI models. Requires an OpenRouter API key.

View on GitHub →

OpenRouter MCP Client for Cursor

A Model Context Protocol (MCP) client for Cursor that utilizes OpenRouter.ai to access multiple AI models.

Requirements

  • Node.js v18.0.0 or later (important!)
  • OpenRouter API key (get one at openrouter.ai/keys)

Features

  • Connect to OpenRouter.ai via MCP
  • Access multiple AI models from various providers (Google, DeepSeek, Meta, etc.)
  • Use MCP transport mechanism to communicate with Cursor
  • Cache model information to reduce API calls
  • Support for both free and paid models
  • Multi-model completion utility to combine results from multiple models

Available Models

This client provides access to all models available on OpenRouter, including:

  • Google Gemini 2.5 Pro
  • DeepSeek Chat v3
  • Meta Llama 3.1
  • DeepSeek R1
  • Qwen Coder
  • Mistral Small 3.1
  • And many more!

Quick Installation

The easiest way to install is to use the setup script:

# Clone the repository
git clone https://your-repo-url/openrouter-mcp-client.git
cd openrouter-mcp-client

# Run the installation script
node install.cjs

The script will:

  1. Help you create a .env file with your OpenRouter API key
  2. Install all dependencies
  3. Build the project
  4. Provide next steps

Manual Installation

If you prefer to install manually:

# Clone the repository
git clone https://your-repo-url/openrouter-mcp-client.git
cd openrouter-mcp-client

# Install dependencies
npm install

# Copy environment file and edit it with your API key
cp .env.example .env
# Edit .env file with your OpenRouter API key

# Build the project
npm run build

Configuration

Edit the .env file with your OpenRouter API key and default model:

OPENROUTER_API_KEY=your_api_key_here
OPENROUTER_DEFAULT_MODEL=google/gemini-2.5-pro-exp-03-25:free

Get your API key from OpenRouter Keys.

Cursor Integration

To use this client with Cursor, you need to update Cursor's MCP configuration file:

  1. Find Cursor's configuration directory:

    • Windows: %USERPROFILE%\.cursor\
    • macOS: ~/.cursor/
    • Linux: ~/.cursor/

  2. Edit or create the mcp.json file in that directory. Add a configuration like this:

{
  "mcpServers": {
    "custom-openrouter-client": {
      "command": "node",
      "args": [
        "FULL_PATH_TO/openrouter-mcp-client/dist/index.js"
      ],
      "env": {
        "OPENROUTER_API_KEY": "your_api_key_here",
        "OPENROUTER_DEFAULT_MODEL": "google/gemini-2.5-pro-exp-03-25:free"
      }
    }
  }
}

Replace FULL_PATH_TO with the actual path to your client installation.

  1. Restart Cursor

  2. Select the client by:

    • Opening Cursor
    • Press Ctrl+Shift+L (Windows/Linux) or Cmd+Shift+L (macOS) to open the model selector
    • Choose "custom-openrouter-client" from the list

Direct Testing (without MCP)

Uncomment the test functions in src/index.ts to test direct API interaction:

// Uncomment to test the direct API
testDirectApi().catch(console.error);
testMultiModelCompletion().catch(console.error);

Then run:

npm start

Development

# Watch mode for development
npm run dev

Troubleshooting

Node.js Version Requirements

Important: This project requires Node.js v18.0.0 or later. If you're using an older version, you will see EBADENGINE warnings and may encounter errors. To check your Node.js version:

node --version

If you have an older version, download and install the latest LTS version from nodejs.org.

Module System Errors

If you encounter errors related to ES modules vs CommonJS:

  • The main codebase uses ES modules (indicated by "type": "module" in package.json)
  • The installation script uses CommonJS (with a .cjs extension)
  • Make sure to run the installation script with node install.cjs

Cursor Not Connecting

If Cursor doesn't seem to connect to your client:

  1. Make sure the path in mcp.json is correct and uses forward slashes
  2. Check that you've built the client with npm run build
  3. Verify that your OpenRouter API key is correct in the env settings
  4. Check Cursor logs for any errors

Related Resources

Smithery Deployment

You can deploy this MCP client to Smithery to make it available to various AI agents and applications.

Prerequisites

  • GitHub account
  • OpenRouter API key

Steps to Deploy

  1. Fork this repository to your GitHub account

  2. Sign in to Smithery at smithery.ai using your GitHub account

  3. Add a new server:

    • Click "Add Server" in the Smithery dashboard
    • Select your forked repository
    • Configure the build settings:
      • Set the base directory to the repository root
      • Ensure the Dockerfile and smithery.yaml are detected

  4. Deploy your server:

    • Smithery will automatically build and deploy your MCP server
    • Once deployed, users can configure it with their own OpenRouter API key

Using Your Deployed Server

After deployment, users can access your server through the Smithery registry:

  1. In their MCP client (like Claude, or any other MCP-compatible client), add the server using Smithery's registry
  2. Configure their OpenRouter API key and preferred default model
  3. Start using the server to access multiple AI models through OpenRouter

Smithery vs. Local Installation

  • Smithery: Easier for distribution to others; no need for users to clone and build the repository
  • Local Installation: Better for personal use and development; more control over the environment

Choose the approach that best fits your needs.

Related Servers