OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
MCP & Tooling

svnscha.de

mcp-windbg is an open-source Model Context Protocol server that connects AI assistants to the Debugging Tools for Windows, allowing them to issue debugger commands and inspect crash dumps or live debugging sessions.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

svnscha.de — OpenViST project profile in MCP & Tooling.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it when an AI assistant needs to interact with WinDbg to investigate a Windows dump or inspect a debugging session. For repeated folder-based triage, the documented Claude Agent SDK example adds separate workflow code around the server.

Who it is for

  • Windows developers investigating application crash dumps
  • Teams building AI-assisted crash-triage workflows
  • Developers inspecting remote or kernel debugging sessions through an AI assistant

Documented uses

  • Issue WinDbg commands through an AI assistant and review the debugger output.
  • Investigate application crash dumps and produce evidence-based triage reports.
  • Compare a new dump’s findings with prior triage reports.
  • Inspect live or remote debugging sessions, including kernel-debugging workflows.
Before you choose

Requirements & limits

The creator describes the project as open source.

  • It is an MCP server connecting an assistant to WinDbg; the debugger executes commands and the model interprets the output.
  • The documented agent example requires Windows, WinDbg, Docker Desktop’s Linux engine, and either Python 3.11+ with uv or Node.js 24.
  • The example uses CDB-oriented tools for application dumps; kernel dumps use KD-oriented tools and need adapted agent instructions.
  • The folder-watching and report-saving workflow is implemented in a separate Claude Agent SDK example, rather than being a built-in server feature.
  • The release notes document both stdio and streamable HTTP transports.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Set up a Claude Agent SDK crash-triage starter

Prepare the documented Python or TypeScript starter project used by the crash-triage workflow.

Before you start

  • Windows with WinDbg
  • Docker Desktop’s Linux engine
  • Either Python 3.11 or later with uv, or Node.js 24
  • The standalone mcp-windbg executable; the guide uses version 1.5.1

Steps

  1. Clone the Python or TypeScript agent-sdk-starter repository.

  2. Install its dependencies with the documented command: the command below for Python or the command below for TypeScript.

    Documented command
    uv sync
    Documented command
    npm ci

  3. Configure .env using that starter repository’s README.

  4. Run the greeting with the command below or the command below.

    Documented command
    uv run --env-file .env agent.py
    Documented command
    npm start

  5. Add the MCP and schema-validation dependencies specified in the guide, then create the triage source file and result schema.

Expected outcome

A starter project is prepared for adding the documented agent-based crash-dump triage workflow.

Guide limitations

  • These steps set up the starter and initial workflow structure; they do not establish that a complete triage run has succeeded.
  • The guide’s example handles user-mode application dumps; kernel dumps require adapting the prompt and allowed commands for bugcheck analysis.
Documentation 1 ↗
Evidence and transparency

Sources & review

svnscha.de ↗Guide documentation ↗
Source record details
Source availability
Active
Sources checked
3 November 2026
Original website title
mcp-windbg - svnscha.de · Official homepage title · retained in its source language
How we review sourcesSuggest a correction