dump-analyzer-mcp-server
MCP server for remote Windows Crash Dump analysis. Enables AI agents to analyze crash dumps via CDB commands through standard MCP interfaces.
README
Dump Analyzer MCP Server
An MCP server for remote Windows Crash Dump analysis. It supports uploading dumps, creating analysis sessions, and executing CDB commands, returning structured status and raw output.
This enables AI agents to perform dump analysis via standard MCP interfaces by invoking CDB deployed on a Windows server, eliminating the need for the AI agent itself to run on a Windows operating system.
<!-- mcp-name: io.github.zuohuiyang/dump-analyzer-mcp-server -->
Core Capabilities
- Wraps the CDB debugger into standard remote MCP calling interfaces
- Provides a complete analysis closed-loop workflow: uploading dumps, creating sessions, executing commands, and closing sessions
Prerequisites
- OS: Windows
- Python: 3.10+
- Debugger: Windows SDK
26100+ (includes WinDbg/CDB) - Clients must be able to access the
--public-base-url
Security Boundaries
- Designed for intranet/trusted environments by default. No built-in authentication. Do not expose directly to the public internet.
- Cross-network access requires authentication, TLS, and access control provided by a frontend gateway.
- Dangerous commands (
.shell, redirection,.create/.attach/.kill, etc.) are rejected by default.
Runtime Parameters
| Parameter | Default | Description |
|---|---|---|
--host |
0.0.0.0 |
Service listen address |
--port |
8000 |
Service listen port |
--public-base-url |
Required | Externally accessible base URL, used to construct and return the dump upload URL to clients |
--cdb-path |
Auto-detect | Path to cdb.exe |
--symbols-path |
srv*c:\symbols*https://msdl.microsoft.com/download/symbols |
Server-side symbol path |
--timeout |
1800 |
Command execution timeout (seconds) |
--upload-dir |
System temp dir | Temporary directory for uploads |
--max-upload-mb |
100 |
Maximum dump file upload size (MB) |
--session-ttl-seconds |
1800 |
Idle session TTL |
--max-active-sessions |
10 |
Maximum active sessions |
--verbose |
false |
Enable DEBUG log level |
Quick Start
Installation via PyPI (Recommended)
pip install dump-analyzer-mcp-server
dump-analyzer-mcp-server --public-base-url http://<your-public-ip-or-domain>:8000
Run via uv (Development)
uv sync
uv run dump-analyzer-mcp-server --public-base-url http://<your-public-ip-or-domain>:8000
- MCP Endpoint:
http://your-host:8000/mcp
MCP Client Configuration
{
"mcpServers": {
"dump-analyzer": {
"url": "http://<your-public-ip-or-domain>:8000/mcp"
}
}
}
Documentation
- Development, Testing, E2E, CI:
docs/development.md - Technical Design & Protocol:
docs/technical-design.md - Development Log:
docs/devlog.md
Acknowledgements
This project evolved from the upstream svnscha/mcp-windbg and has been focused as a dedicated MCP server for remote Windows Crash Dump analysis.
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。