Skip to main content
Visual Studio Code and GitHub Copilot can act as MCP clients. Once you point VS Code at your server, every tool it exposes becomes available through the Copilot Chat interface — controlled by natural language, without writing any additional client code.

​
Learning objectives

By the end of this lesson you will be able to:
  • Configure VS Code to find and connect to your MCP server
  • Discover server tools through the Copilot Chat tools panel
  • Invoke tools via natural-language prompts in Agent mode
  • Understand how to pass secrets like API keys to MCP servers

​
Two ways to control VS Code MCP integration

Visual interface

Edit .vscode/mcp.json and use the play button in the editor to start/stop servers.

Terminal / CLI

Use the code --add-mcp command to add a server directly from the shell:
code --add-mcp '{"name":"my-server","command":"uvx","args":["mcp-server-fetch"]}'

​
Exercise: Consuming a server from VS Code

1

Enable MCP server discovery

  1. Open File → Preferences → Settings (or Ctrl+, / Cmd+,).
  2. Search for MCP.
  3. Enable chat.mcp.discovery.enabled in your settings.json.
2

Create the config file

In your project root, create a .vscode directory and add an mcp.json file:
.vscode/
└── mcp.json
3

Configure your server

Add a server entry to .vscode/mcp.json. The command and args tell VS Code how to start the server process.
{
  "inputs": [],
  "servers": {
    "hello-mcp": {
      "command": "node",
      "args": ["build/index.js"]
    }
  }
}
For a Python server:
{
  "inputs": [],
  "servers": {
    "calculator-py": {
      "command": "python",
      "args": ["server.py"]
    }
  }
}
To prompt the user for an API key at startup, add an input:
{
  "inputs": [
    {
      "type": "promptString",
      "id": "weather-api-key",
      "description": "Weather API Key",
      "password": true
    }
  ],
  "servers": {
    "weather-sse": {
      "type": "sse",
      "url": "http://localhost:8000/sse",
      "headers": {
        "x-api-key": "${input:weather-api-key}"
      }
    }
  }
}
4

Start the server

Open mcp.json in the VS Code editor. Locate the play icon (▶) next to your server entry and click it. The server process starts and VS Code connects to it.Once connected, the tools icon in the GitHub Copilot Chat panel shows the number of available tools. Click it to see the list and toggle individual tools on or off.
5

Invoke a tool with a prompt

Open the GitHub Copilot Chat panel (Ctrl+Shift+I / Cmd+Shift+I), switch to Agent mode, and type a prompt that matches one of your server’s tools:
add 22 to 1
Copilot should respond with 23 by calling the add tool on your server.
You can prompt users for inputs like API keys directly in mcp.json using inputs with type: "promptString". VS Code will show a secure input dialog before starting the server.

​
Assignment

Add a server entry to your mcp.json file and confirm you can start and stop it. Then test communication with your server’s tools through the GitHub Copilot Chat interface.

​
Key takeaways

  • VS Code is a powerful MCP host that lets you consume multiple MCP servers without writing client code.
  • GitHub Copilot Chat’s Agent mode is the primary interface for interacting with server tools.
  • You can prompt users for inputs like API keys when configuring the server entry in mcp.json.
  • The code --add-mcp command lets you register servers directly from the terminal.

Build docs developers (and LLMs) love

Get started for freeTalk to us