Win MCP Server
Enables AI agents to securely manage and execute commands on remote Windows servers via WinRM, including PowerShell execution, system information retrieval, and service management.
README
Win MCP Server
A Model Context Protocol (MCP) server that provides AI agents with the ability to interact with Windows servers using Windows Remote Management (WinRM). This server enables secure remote PowerShell execution and system management tasks on Windows hosts.
Features
- Secure Authentication: Uses domain credentials stored in macOS Keychain with TouchID authentication
- PowerShell Execution: Execute arbitrary PowerShell commands on remote Windows hosts
- System Information: Get basic system information, running services, and disk space
- Error Handling: Comprehensive error handling with helpful error messages
- Memory Safety: Credentials are cleared from memory immediately after use
Installation
From PyPI (Recommended)
pip install win-mcp-server
From Source
git clone https://github.com/rorymcmahon/win-mcp-server.git
cd win-mcp-server
pip install -e .
Configuration
Prerequisites
-
WinRM Configuration: Ensure target Windows hosts have WinRM enabled and configured.
-
Credential Setup: The server will prompt for credentials on first use and cache them securely in macOS Keychain with 4-hour expiration.
MCP Configuration
Add to your MCP settings file (e.g., ~/.config/mcp/settings.json):
{
"mcpServers": {
"winrm": {
"command": "winrm-mcp-server"
}
}
}
Available Tools
setup_credentials(hostname: str)
Setup credentials for a Windows host (interactive mode).
Parameters:
hostname: The target Windows hostname (FQDN)
Returns:
status: Success or error statusmessage: Confirmation message
execute_powershell(hostname: str, command: str)
Execute arbitrary PowerShell commands on a remote Windows host.
Parameters:
hostname: The target Windows hostname (FQDN)command: PowerShell command to execute
Returns:
status: Exit code of the commandstdout: Standard output from the commandstderr: Standard error from the command
get_system_info(hostname: str)
Get basic system information from a Windows host.
Returns: JSON with Windows product name, total physical memory, and processor information.
get_running_services(hostname: str)
Get list of running services from a Windows host.
Returns: JSON array of running services with name, status, and start type.
get_disk_space(hostname: str)
Get disk space information from a Windows host.
Returns: JSON array of logical disks with device ID, total size, and free space in GB.
Usage Examples
Initial Setup
"Setup credentials for server01.domain.local"
Basic System Check
"Get system info and disk space for server01.domain.local"
Service Management
"Check if IIS is running on webserver.domain.local"
"Get running services on server01.domain.local"
Custom PowerShell Commands
"Run 'Get-EventLog -LogName System -Newest 10' on server01.domain.local"
"Execute 'Get-Process | Sort-Object CPU -Descending | Select-Object -First 10' on server01.domain.local"
Security
- Credential Storage: Domain credentials are securely stored in macOS Keychain
- TouchID Authentication: Credentials require TouchID authentication for access
- Memory Safety: Passwords are immediately cleared from memory after use
- Transport Security: Uses NTLM authentication over HTTP (configurable for HTTPS)
- Automatic Expiration: Cached credentials expire after 4 hours
- Secure Prompting: Password input is hidden and never logged
Development
Setup Development Environment
git clone https://github.com/rorymcmahon/winrm-mcp-server.git
cd winrm-mcp-server
pip install -e ".[dev]"
Running Tests
pytest
Code Formatting
black src/
isort src/
Requirements
- Python 3.10+
- macOS (for Keychain integration)
- Target Windows hosts with WinRM enabled
License
MIT License - see LICENSE file for details.
Contributing
- Fork the repository
- Create a feature branch
- Make your changes
- Add tests if applicable
- Submit a pull request
Support
For issues and questions:
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。