Skip to content

getting-started.md

Latest commit

 

History

History
105 lines (75 loc) · 2.83 KB

getting-started.md

File metadata and controls

105 lines (75 loc) · 2.83 KB

Getting Started with Debug MCP Server

This guide will walk you through testing the Debug MCP Server locally with a simple Python example. The server also supports JavaScript, Rust, Go, Java, and .NET/C# debugging -- see the language-specific guides for details.

Prerequisites

  1. Make sure you've completed installation:

    pnpm install
npm run build

  2. Check that Python and debugpy are installed:

    python --version
pip list | grep debugpy

  3. Verify the MCP settings are configured properly in VS Code

Step-by-Step Testing

1. Start Claude VS Code Extension

Open VS Code and ensure the Claude extension is running and connected.

2. Test the Example Python Script

The repository includes a simple Fibonacci calculator in examples/python/fibonacci.py. Let's debug this file.

Using Claude to Debug

In a new conversation with Claude, try these prompts:

  1. Create a debug session:

    Create a new Python debug session named "Fibonacci Test"

  2. Set a breakpoint:

    Set a breakpoint in examples/python/fibonacci.py at line 21

    (Line 21 is inside the fibonacci_iterative function)

  3. Start debugging:

    Start debugging examples/python/fibonacci.py

  4. Step through the code:

    Step over the current line

    Or

    Step into the function call

  5. Inspect variables:

    Show me all the variables in the current scope

  6. Evaluate an expression:

    Evaluate n + 1 in the current context

  7. Continue execution:

    Continue execution to the next breakpoint

  8. Close the session when finished:

    Close the debug session

Checking Server Status

If you encounter issues, you can check the server status in VS Code:

  1. Click on the Claude extension icon in the VS Code sidebar
  2. Look for the "debug-mcp-server" entry in the MCP Servers list
  3. Check if it shows as "Connected" or if there are any error messages

Understanding the Server Logs

The server suppresses all console output in all transport modes (stdio, SSE, etc.) to avoid corrupting the JSON-RPC protocol stream. Logs are only written when a --log-file path is specified. To inspect logs:

  1. Configure a log file in your MCP settings: "args": ["dist/index.js", "stdio", "--log-file", "/path/to/debug.log"]
  2. Check the log file for "error" level entries
  3. Look for messages about Python detection and debugpy availability
  4. Monitor DAP (Debug Adapter Protocol) communication logs

Next Steps

Once you've verified the server works with the example, you can try:

  1. Debugging your own Python scripts
  2. Exploring more complex debugging scenarios
  3. Testing the source code viewing and variable inspection features

For more details on available commands, see the Usage Guide.