Getting Started with MCP in Cursor

Prerequisites

Before setting up MCP in Cursor, make sure you have the following:

• Cursor version 0.45 or later (MCP support was added in early 2025)
• Node.js version 18 or higher (required for running MCP servers via npx)
• A project directory you want the AI to access

You can verify your Node.js version from the terminal:

node --version
# Expected output: v18.x.x or higher

If you need to install or update Node.js, use a version manager like nvm or download the latest LTS release from the official Node.js website.

Installing Your First Server

Cursor reads MCP server configuration from a .cursor/mcp.json file in your project root. Create this file with a filesystem server entry:

{
  "mcpServers": {
    "filesystem": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-filesystem",
        "/Users/you/projects/my-app"
      ]
    }
  }
}

Replace /Users/you/projects/my-app with the absolute path to your project directory. The -y flag tells npx to automatically confirm the package download on first run.

After saving the file, restart Cursor or reload the window. Cursor will detect the configuration and start the MCP server automatically.

Testing the Connection

Once the server is running, open Cursor's AI chat and try a request that exercises the filesystem tools:

1. Ask the AI to list files in your project directory.
2. Ask it to read a specific file and summarize its contents.
3. Ask it to search for a pattern across your codebase.

If the connection is working, the AI will use the MCP filesystem tools to fulfill these requests directly. You should see tool call indicators in the chat showing which MCP operations were performed.

If the server fails to start, check the Cursor output panel for errors. The most common issues are:

• Wrong Node.js version — the server package requires Node 18+.
• Invalid JSON — run your config through a linter with cat .cursor/mcp.json | jq .
• Path does not exist — verify the directory path in your args array.

Adding More Servers

The real power of MCP comes from combining multiple servers. You can add a GitHub server alongside your filesystem server to give Cursor access to issues, pull requests, and repository metadata:

{
  "mcpServers": {
    "filesystem": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-filesystem", "/Users/you/projects/my-app"]
    },
    "github": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-github"],
      "env": {
        "GITHUB_PERSONAL_ACCESS_TOKEN": "ghp_your_token_here"
      }
    }
  }
}

Each server runs as an independent process. Cursor discovers the tools each server provides and makes them available to the AI agent. You can add as many servers as your workflow requires — databases, APIs, cloud services, and custom tools all work through the same protocol.

Start with one server, verify it works, and then expand. This incremental approach makes troubleshooting much simpler than configuring everything at once.