Connecting to Cursor & VS Code

Configure your MCP servers to work with Cursor IDE and VS Code. Learn the configuration formats, project-level vs global setup, and how to use MCP tools in your coding workflow.

title: "Connecting to Cursor & VS Code" description: "Configure your MCP servers to work with Cursor IDE and VS Code. Learn the configuration formats, project-level vs global setup, and how to use MCP tools in your coding workflow." order: 8 level: "beginner" duration: "10 min" keywords:

• Cursor MCP setup
• VS Code MCP
• MCP IDE integration
• Cursor MCP configuration
• VS Code MCP server
• MCP code editor
• Cursor AI tools
• mcp.json configuration date: "2026-04-01"

Why Connect MCP Servers to Your Code Editor?

Connecting MCP servers to your code editor brings AI-powered tools directly into your development workflow. Instead of switching between applications, you can:

• Query databases while writing code
• Access internal documentation without leaving the editor
• Run custom analysis tools on your codebase
• Use prompt templates for code review and documentation
• Interact with APIs and external services through natural language

How Do You Configure MCP in Cursor?

Cursor has built-in MCP support. You can configure MCP servers at the project level or globally.

Project-Level Configuration

Create a .cursor/mcp.json file in your project root:

{
  "mcpServers": {
    "weather-server": {
      "command": "node",
      "args": ["/absolute/path/to/weather-server/dist/index.js"]
    }
  }
}

Project-level configuration is ideal when an MCP server is specific to a particular project — for example, a database server configured for that project's database.

Global Configuration

For MCP servers you want available across all projects, add them to Cursor's global settings:

1. Open Cursor Settings (Cmd/Ctrl + Shift + J, or via the gear icon)
2. Navigate to the MCP section
3. Add your server configuration

Alternatively, edit ~/.cursor/mcp.json:

{
  "mcpServers": {
    "github-tools": {
      "command": "node",
      "args": ["/path/to/github-tools/dist/index.js"],
      "env": {
        "GITHUB_TOKEN": "ghp_your_token"
      }
    },
    "general-utils": {
      "command": "node",
      "args": ["/path/to/general-utils/dist/index.js"]
    }
  }
}

Complete Cursor Setup Walkthrough

cd /path/to/your/mcp-server
npm run build

mkdir -p .cursor

{
  "mcpServers": {
    "my-server": {
      "command": "node",
      "args": ["/absolute/path/to/my-server/dist/index.js"]
    }
  }
}

Cursor Configuration Options

Cursor supports the same configuration fields as Claude Desktop:

{
  "mcpServers": {
    "server-name": {
      "command": "node",
      "args": ["/path/to/server/dist/index.js"],
      "env": {
        "API_KEY": "your-key",
        "DATABASE_URL": "your-db-url"
      }
    }
  }
}

You can also use npx for published servers:

{
  "mcpServers": {
    "published-server": {
      "command": "npx",
      "args": ["-y", "some-mcp-server-package"]
    }
  }
}

How Do You Configure MCP in VS Code?

VS Code supports MCP servers through its built-in agent mode and GitHub Copilot integration. The configuration uses a slightly different format.

Project-Level Configuration

Create a .vscode/mcp.json file in your project root:

{
  "servers": {
    "weather-server": {
      "type": "stdio",
      "command": "node",
      "args": ["/absolute/path/to/weather-server/dist/index.js"]
    }
  }
}

Note the key differences from Cursor's format:

• The top-level key is servers (not mcpServers)
• Each server has a type field (typically "stdio")

User-Level (Global) Configuration

For servers available across all VS Code projects, add them to your user settings (settings.json):

{
  "mcp": {
    "servers": {
      "github-tools": {
        "type": "stdio",
        "command": "node",
        "args": ["/path/to/github-tools/dist/index.js"],
        "env": {
          "GITHUB_TOKEN": "ghp_your_token"
        }
      }
    }
  }
}

Access your user settings with Cmd/Ctrl + Shift + P and search for "Preferences: Open User Settings (JSON)".

Complete VS Code Setup Walkthrough

cd /path/to/your/mcp-server
npm run build

{
  "servers": {
    "my-server": {
      "type": "stdio",
      "command": "node",
      "args": ["/absolute/path/to/my-server/dist/index.js"]
    }
  }
}

Configuration Comparison Across Editors

Setting Up a Server for All Three Clients

If you want your MCP server to work with Claude Desktop, Cursor, and VS Code simultaneously, you need to configure it in each client separately. The server code does not change — only the client configuration differs.

Here is a script to help you set up all three configurations for a given server:

#!/bin/bash
# setup-mcp-server.sh
# Usage: ./setup-mcp-server.sh /path/to/server server-name

SERVER_PATH="$1"
SERVER_NAME="$2"

echo "Server: $SERVER_NAME"
echo "Path: $SERVER_PATH/dist/index.js"
echo ""
echo "=== Claude Desktop ==="
echo "Add to ~/Library/Application Support/Claude/claude_desktop_config.json:"
echo ""
echo "\"$SERVER_NAME\": {"
echo "  \"command\": \"node\","
echo "  \"args\": [\"$SERVER_PATH/dist/index.js\"]"
echo "}"
echo ""
echo "=== Cursor ==="
echo "Add to .cursor/mcp.json:"
echo ""
echo "\"$SERVER_NAME\": {"
echo "  \"command\": \"node\","
echo "  \"args\": [\"$SERVER_PATH/dist/index.js\"]"
echo "}"
echo ""
echo "=== VS Code ==="
echo "Add to .vscode/mcp.json:"
echo ""
echo "\"$SERVER_NAME\": {"
echo "  \"type\": \"stdio\","
echo "  \"command\": \"node\","
echo "  \"args\": [\"$SERVER_PATH/dist/index.js\"]"
echo "}"

Using MCP Servers Effectively in Code Editors

In Cursor

Cursor provides two main interfaces for using MCP tools:

• Chat (Cmd/Ctrl + L) — Conversational interface where you can ask questions and Cursor will use MCP tools as needed
• Composer (Cmd/Ctrl + I) — Inline editing assistant that can leverage MCP tools while modifying your code

When you ask Cursor a question, it automatically evaluates which MCP tools are relevant and calls them. You can also explicitly request a tool be used.

In VS Code

VS Code's Copilot Chat uses MCP tools in agent mode. Use the chat panel and prefix with @workspace or use the agent mode to access MCP capabilities.

Project-Level Configuration Best Practices

Example .gitignore Entry

# Ignore MCP configs with local paths
.cursor/mcp.json
.vscode/mcp.json

Example Shareable Configuration (Using npx)

If your MCP server is published to npm, you can share the configuration:

{
  "mcpServers": {
    "team-tools": {
      "command": "npx",
      "args": ["-y", "@your-org/team-mcp-tools"],
      "env": {
        "API_KEY": "${TEAM_API_KEY}"
      }
    }
  }
}

Troubleshooting Editor Connections

Server Not Appearing

1. Verify the path — Use absolute paths, not relative
2. Check file location — .cursor/mcp.json must be in the project root
3. Validate JSON — Even a single missing comma breaks the config
4. Reload the editor — Configuration changes require a reload
5. Check the server builds — Run node /path/to/dist/index.js directly to verify it starts

Tools Not Being Called

• The AI may not always choose to use your tools — be explicit in your requests
• Check that tool names and descriptions are clear
• Verify the server is showing as connected in the editor's MCP settings
• Try asking specifically: "Use the get_weather tool to check Tokyo"

Slow Response Times

• Check if your tool's execute method has long-running operations
• Add timeouts to external API calls
• Consider caching results for frequently accessed data
• Monitor the server process for memory leaks

What You Have Learned in the Beginner Path

Congratulations on completing the beginner learning path. Here is a recap of everything you have covered:

What Comes Next?

You now have the foundation to build real MCP servers. Here are some paths to explore:

• Intermediate Path — Dive deeper into advanced tool patterns, dynamic resources, transport protocols, and testing
• Build a Real Server — Pick an API or service you use daily and build an MCP server for it
• Explore the Ecosystem — Browse existing MCP servers for inspiration and to learn patterns from other developers
• Publish Your Server — Share your MCP server on npm so others can use it with a simple npx command

You have completed the beginner learning path. You now have the skills to build, configure, and deploy MCP servers with any major AI client. Start building something real — the best way to learn is by doing.