dap-mcp

Name: dap-mcp
Author: KashunCheng

@KashunCheng

Repository

About dap-mcp

Model Context Protocol (MCP) server that interacts with a Debugger

#debugging #mcp #debug #dap #debug-adapter-protocol #llm #model-context-protocol

Basic information

Config

Add this server to your MCP-compatible client using the configuration below.

{
  "mcpServers": {
    "dap_mcp": {
      "command": "python",
      "args": [
        "-m",
        "dap_mcp",
        "--config",
        "config.json"
      ]
    }
  }
}

Tools

** Launch the debuggee program.

** Set a breakpoint at a specified file and line (with an optional condition).

** Remove a breakpoint from a specified file and line.

** List all breakpoints currently set in the debugger.

** Continue program execution after hitting a breakpoint.

** Step into a function call.

** Step out of the current function.

** Step over to the next line of code.

** Evaluate an expression in the current debugging context.

** Switch to a different stack frame.

** View source code around a specified line (using the last provided file if none is specified).

** Terminate the debugging session.

Overview

What is dap-mcp?

dap-mcp is an MCP server that integrates the Model Context Protocol (MCP) with the Debug Adapter Protocol (DAP) to manage debugging sessions. It is designed for developers who want to leverage LLMs to streamline debugging workflows.

How to use dap-mcp?

Install with pip install dap-mcp, then run with python -m dap_mcp --config config.json. Optionally use uvx dap-mcp@latest --config config.json. Configure debugger settings, source directories, and debuggee parameters via a JSON configuration file.

Key features of dap-mcp

Manage Debug Adapter Protocol (DAP) sessions.
Set, list, and remove breakpoints.
Control execution (continue, step in/out/next).
Evaluate expressions in current debugging context.
Switch stack frames and view source code.
Flexible JSON configuration for debugger settings.

Use cases of dap-mcp

Debug a Python program using debugpy with LLM-powered breakpoint management.
Step through code interactively while an AI assistant inspects variable values.
Evaluate expressions in a live debugging session from an MCP client.
Automate breakpoint toggling across multiple source files during debugging.

FAQ from dap-mcp

What protocols does dap-mcp support?

dap-mcp integrates the Debug Adapter Protocol (DAP) for debugger communication and the Model Context Protocol (MCP) for AI-assisted context management.

What are the runtime requirements?

Python 3.10 or higher is required. The uv package manager is optional but recommended for running the server.

Where is the configuration stored?

Configuration is stored in a JSON file (e.g., config.json) that specifies the debugger path, source directories, module, working directory, and interpreter.

Can I add support for additional debuggers?

Yes. Add a new DAP configuration class in dap_mcp/config.py by subclassing DAPConfig with a unique type literal, then update the union type for validation.