ProxmoxEmCP
Enables AI assistants to manage Proxmox VE infrastructure, including VMs, containers, storage, and cluster operations via natural language.
README
Proxmox MCP Server
A Model Context Protocol (MCP) server for managing Proxmox VE infrastructure through AI assistants. Available as an npm package, Docker container, or standalone Python application.
Project Background
In May 2025, we launched the original ProxmoxMCP server with FastMCP, FastAPI, and virtual environments. Based on community feedback and operational experience, we rebuilt the project as ProxmoxEmCP - a cleaner, simpler implementation that eliminates setup complexity while maintaining full functionality. This version uses the official MCP SDK directly and runs without virtual environment dependencies, making deployment and maintenance significantly easier.
Table of Contents
- Quick Start
- Prerequisites
- Installation Methods
- Configuration
- AI Agent Integration
- Available MCP Tools
- Architecture
- Development
- Troubleshooting
- License
- Support
Quick Start
npm/npx
# Install globally
npm install -g @puregrain/proxmox-emcp-node
# Or run directly with npx
npx @puregrain/proxmox-emcp-node
Docker Hub
docker run -d \
--name proxmox-emcp \
-e PROXMOX_HOST="192.168.1.100" \
-e PROXMOX_TOKEN_ID="your-token-id" \
-e PROXMOX_TOKEN_SECRET="your-token-secret" \
puregrain/proxmox-emcp:latest
Prerequisites
- Proxmox VE: Version 7.0 or higher with API access enabled
- For npm package: Node.js 18+
- For Docker: Docker Engine installed
- For local Python: Python 3.9+
- API Token: Created in Proxmox with appropriate permissions
Installation Methods
npm Package
The npm package provides a native Node.js implementation without Docker requirements.
# Install globally
npm install -g @puregrain/proxmox-emcp-node
# Set environment variables
export PROXMOX_HOST="192.168.1.100"
export PROXMOX_TOKEN_ID="your-token-id"
export PROXMOX_TOKEN_SECRET="your-token-secret"
# Run the server
proxmox-emcp-node
Using with npx (no installation required):
PROXMOX_HOST=192.168.1.100 \
PROXMOX_TOKEN_ID=your-token-id \
PROXMOX_TOKEN_SECRET=your-token-secret \
npx @puregrain/proxmox-emcp-node
Docker Container
Using Docker Compose (Recommended)
- Clone the repository:
git clone https://github.com/PureGrain/ProxmoxEmCP.git
cd ProxmoxEmCP
- Configure environment:
cp .env.example .env
# Edit .env with your Proxmox credentials
- Start the container:
docker-compose up -d
Using Docker CLI
docker run -d \
--name proxmox-emcp \
-e PROXMOX_HOST="192.168.1.100" \
-e PROXMOX_TOKEN_ID="your-token-id" \
-e PROXMOX_TOKEN_SECRET="your-token-secret" \
puregrain/proxmox-emcp:latest
Using GitHub Container Registry
docker run -d \
--name proxmox-emcp \
-e PROXMOX_HOST="192.168.1.100" \
-e PROXMOX_TOKEN_ID="your-token-id" \
-e PROXMOX_TOKEN_SECRET="your-token-secret" \
ghcr.io/puregrain/proxmox-emcp:latest
Local Python
# Clone the repository
git clone https://github.com/PureGrain/ProxmoxEmCP.git
cd ProxmoxEmCP
# Install dependencies
pip install -r requirements.txt
# Set environment variables
export PROXMOX_HOST="192.168.1.100"
export PROXMOX_TOKEN_ID="your-token-id"
export PROXMOX_TOKEN_SECRET="your-token-secret"
# Run the server
python mcp_server_stdio.py
Open WebUI Integration
For Open WebUI users, we provide ProxmoxWeaver - a specialized tool that brings full Proxmox management capabilities directly into your Open WebUI interface.
ProxmoxWeaver is a native Open WebUI tool that provides the same comprehensive Proxmox VE management features as our MCP server, but specifically designed for the Open WebUI ecosystem.
Features:
- Direct integration with Open WebUI's tool system
- Same powerful Proxmox management capabilities
- No additional containers or services required
- Simple installation through Open WebUI's tool marketplace
Installation:
- Visit the ProxmoxWeaver repository
- Copy the tool configuration
- In Open WebUI, navigate to Tools → Add Tool
- Paste the ProxmoxWeaver configuration
- Configure your Proxmox credentials in the tool settings
Usage:
Once installed, ProxmoxWeaver appears as a native tool in Open WebUI, allowing you to:
- Query and manage VMs and containers
- Monitor cluster health and resources
- Execute commands and create snapshots
- Manage storage, backups, and templates
- All through natural language interactions in Open WebUI
This integration is perfect for teams already using Open WebUI who want to add Proxmox management capabilities without additional infrastructure.
Configuration
Environment Variables
| Variable | Required | Default | Description |
|---|---|---|---|
PROXMOX_HOST |
Yes | - | Proxmox server IP or hostname |
PROXMOX_TOKEN_ID |
Yes | - | API token ID |
PROXMOX_TOKEN_SECRET |
Yes | - | API token secret |
PROXMOX_USER |
No | root@pam | Proxmox user |
PROXMOX_VERIFY_SSL |
No | false | Verify SSL certificates |
LOG_LEVEL |
No | INFO | Logging level (DEBUG, INFO, WARNING, ERROR) |
Creating API Token
- Log into Proxmox Web UI
- Navigate to Datacenter → Permissions → API Tokens
- Click Add to create a new token
- Configure the token:
- User: Select your user (e.g., root@pam)
- Token ID: Choose a descriptive name
- Privilege Separation: Uncheck for full user permissions
- Copy the token secret (shown only once!)
AI Agent Integration
Configure your AI assistant (Claude Desktop, Cline, or any MCP-compatible client) to connect to the server.
For npm Package
{
"mcpServers": {
"proxmox": {
"command": "npx",
"args": ["@puregrain/proxmox-emcp-node"],
"env": {
"PROXMOX_HOST": "192.168.1.100",
"PROXMOX_TOKEN_ID": "your-token-id",
"PROXMOX_TOKEN_SECRET": "your-token-secret"
}
}
}
}
For Docker Container
{
"mcpServers": {
"proxmox": {
"command": "docker",
"args": ["attach", "proxmox-emcp"],
"env": {
"PROXMOX_HOST": "192.168.1.100",
"PROXMOX_TOKEN_ID": "your-token-id",
"PROXMOX_TOKEN_SECRET": "your-token-secret"
}
}
}
}
Available MCP Tools
Node Management
get_nodes- List all nodes in the clusterget_node_status- Get detailed status for a specific node
VM Operations
get_vms- List all VMs across the clusterget_vm_status- Get VM status and configurationstart_vm- Start a virtual machinestop_vm- Stop a virtual machine gracefullyreboot_vm- Reboot a virtual machineexecute_vm_command- Execute commands via QEMU guest agentcreate_vm_snapshot- Create a VM snapshotlist_vm_snapshots- List all snapshots for a VMget_vm_network- Get VM network configuration
Container Operations (LXC)
get_containers- List all LXC containersget_container_status- Get container status and configurationstart_container- Start a containerstop_container- Stop a container gracefullyreboot_container- Reboot a containerexecute_container_command- Execute commands in containercreate_container_snapshot- Create a container snapshotlist_container_snapshots- List container snapshots
Storage & Backup
get_storage- List storage poolsget_storage_details- Get detailed storage pool informationget_backups- List backup files with filteringlist_templates- List VM and container templates
Monitoring & Logs
get_cluster_status- Get comprehensive cluster health and resourcesget_task_status- Check Proxmox task statusget_recent_tasks- List recent tasks with filteringget_cluster_log- Get cluster-wide log entries
Access Control
get_users- List all users with groups and tokensget_groups- List all groups with membersget_roles- List all roles and privileges
Network & Security
get_firewall_status- Get firewall status and rules
Architecture
┌─────────────────┐
│ AI Assistant │
│ (Claude, etc) │
└────────┬────────┘
│ MCP Protocol
▼
┌─────────────────┐
│ MCP Server │
│ (Node/Python) │
└────────┬────────┘
│ REST API
▼
┌─────────────────┐
│ Proxmox VE │
│ Cluster │
└─────────────────┘
Development
Contributing
- Fork the repository
- Create a feature branch
- Make your changes
- Run tests:
npm testorpython -m pytest - Submit a pull request
Pre-commit Hooks
This project uses pre-commit for code quality:
pip install pre-commit
pre-commit install
pre-commit run --all-files
Hooks include:
black- Python code formatterflake8- Python linterdetect-secrets- Secret scanning- YAML and whitespace validation
Building Docker Image
docker build -t proxmox-emcp .
Modifying the Server
- Node.js version: Edit files in
npm-app/ - Python version: Edit
mcp_server_stdio.py - Docker config: Edit
Dockerfileanddocker-compose.yml
Troubleshooting
Connection Issues
- Verify Proxmox is reachable:
ping <PROXMOX_HOST> - Check API port: Ensure port 8006 is accessible
- SSL certificates: Set
PROXMOX_VERIFY_SSL=falsefor self-signed certs
Authentication Failures
- Token permissions: Ensure token has required permissions
- Token format: Verify
TOKEN_IDandTOKEN_SECRETare correct - User privileges: Check user has appropriate Proxmox permissions
Viewing Logs
# Docker logs
docker logs proxmox-emcp
# npm/Node.js - set LOG_LEVEL
LOG_LEVEL=DEBUG npx @puregrain/proxmox-emcp-node
Common Issues
- "Connection refused": Check firewall and Proxmox API service
- "Unauthorized": Verify token credentials
- "SSL verification failed": Set
PROXMOX_VERIFY_SSL=false
License
This project is dual-licensed under the following terms:
- MIT License: For open-source use. See the LICENSE file for details.
- Commercial License: For proprietary use. See the COMMERCIAL-LICENSE.txt file for details.
For commercial licensing inquiries, please contact: [Your Email or Website]
Support
- Issues: GitHub Issues
- Documentation: MCP Configuration Guide
- Sponsor: GitHub Sponsors
Author: PureGrain at SLA Ops, LLC Repository: github.com/PureGrain/ProxmoxEmCP
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。