MCP Server for WinDbg Crash Analysis

You can run an MCP server that connects AI models with WinDbg to analyze Windows crash dumps and engage in live debugging. This server wraps around CDB and exposes a consistent MCP interface so your AI assistant can issue debugger commands, inspect dumps, and work with remote debugging targets through familiar clients.

How to use

You will interact with the MCP Windbg server through an MCP-enabled client. Start the server with a local transport for quick experiments or use an HTTP transport for remote access. Typical workflows include analyzing a crash dump, connecting to a remote debugging session, and processing multiple dumps to identify recurring patterns.

How to install

Prerequisites you need before installing the server:

• Windows with Debugging Tools for Windows or WinDbg from Microsoft Store

• Python 3.10 or higher

• Any MCP-compatible client (GitHub Copilot, Claude Desktop, Cline, Cursor, Windsurf, etc.)

Install the MCP Windbg server from PyPI and start using it in your environment.

pip install mcp-windbg

Available tools

list_windbg_dumps

List crash dump files available for analysis to quickly discover and select targets for investigation.

open_windbg_dump

Open and analyze a selected crash dump to perform initial investigations and extract relevant data.

close_windbg_dump

Close and clean up resources associated with an opened crash dump.

open_windbg_remote

Connect to a remote debugging target to perform live debugging across machines.

close_windbg_remote

Terminate remote debugging sessions and free associated resources.

run_windbg_cmd

Execute WinDbg commands or sequences to perform custom analysis and investigation.