VS Code Debugger MCP Server

Give your AI coding assistant debugging superpowers in Visual Studio Code.

Why Install This Extension?

Modern AI assistants can write code, but they can't debug it—until now. This extension acts as a secure bridge between your favorite AI tools (like GitHub Copilot, Roo Code, Claude, and others) and the VS Code debugger. With it, your AI can:

• Run and debug your applications in any language supported by VS Code.
• Set breakpoints to pause execution at specific lines.
• Inspect variables and the call stack to understand your code's state.
• Step through your code line-by-line to trace logic and find errors.

No more cluttered logs or messy debug code — let your AI handle it smartly inside VS Code.

Debug smarter, not louder.

💖 Like this extension?

Support the development here:
👉 paypal.me/rauschit

Every donation helps keep the tools free and evolving.

👤 Free for personal use
💼 Commercial use requires a license
📩 Get in touch: license@rauschit.de

How It Works

This extension runs quietly in the background as a secure MCP (Model Context Protocol) server. Most modern AI assistants and development tools support MCP and can connect directly to this server.

When your AI assistant needs to start a debugging session, it sends a request via MCP. The extension then translates that request into commands for VS Code’s built-in debugger—all behind the scenes.

As a user, you simply tell your AI assistant something like: "Debug this script" or "Find out why this function crashes." The AI takes it from there—starting the debugger, inspecting stack traces, and identifying problematic variables—without you having to write a single log or breakpoint.

Getting Started in 3 Steps

1. Install This Extension

Install VS Code Debugger MCP Server from the VS Code Marketplace.

2. Configure This Extension

1. Open VS Code's Settings (Ctrl+,).
2. Go to Extensions > VSCode Debugger MCP Server.
3. Choose a Transport Type (streamable-http or sse) and Port. The default port 10101 works for most users.

Note: If your AI tool requires stdio, you may need to run the provided stdio proxy app (see documentation/examples).

3. Configure Your AI Assistant

Tell your AI assistant how to connect to this server. Below are configuration examples for common tools and transport types. Add the appropriate snippet to your AI tool's MCP/server configuration.

Example MCP Client Configurations

Streamable HTTP MCP Server Config for Roo Code

"mcp.servers": {
  "vscode-debugger": {
    "type": "streamable-http",
    "url": "http://localhost:10101/mcp"
  }
}

Streamable HTTP MCP Server Config for Github Copilot

{
	"servers": {
		"VS Code Debugger MCP Server": {
			"url": "http://localhost:10101/mcp",
			"type": "http"
		}
	},
	"inputs": []
}

SSE MCP Server Config for Roo Code

{
  "mcpServers": {
    "vscode-debugger": {
      "type": "sse",
      "url": "http://localhost:10101/mcp-events",
      "description": "VSCode Debugger MCP Server"
    }
  }
}

Supported Debugging Tools (MCP Commands)

• start: Start a new debug session.
• stop: Stop the current debug session.
• pause: Pause execution.
• continue: Continue execution.
• step_over, step_in, step_out: Step through code.
• set_breakpoints, set_breakpoint, remove_breakpoint, clear_all_breakpoints, get_breakpoint, get_all_breakpoints, toggle_breakpoint: Breakpoint management.
• get_stack_trace, get_threads, get_scopes, get_variables: Inspect program state.
• evaluate_expression: Evaluate code in the current context.
• add_watch, remove_watch, list_watches, update_all_watches, edit_watch: Manage watch expressions.

License

SEE LICENSE IN LICENSE FILE