podman-mcp
Enables management of Podman containers, pods, images, and compose stacks via natural language, with support for container stats, logs, exec, health analysis, and a web dashboard.
README
🦭 Podman MCP (podman-mcp)
A premium, daemon-less FastMCP 3.5 control plane for Podman container engines on Windows and WSL2 — featuring container management, native Pod orchestration, compose stacks, system analytics, and a glassmorphism React web dashboard.
❓ What is Podman?
Podman (Pod Manager) is a lightweight, open-source container engine designed as a modern alternative to Docker.
While Docker relies on a monolithic background service (the Docker Daemon, dockerd) which can occasionally lock up or consume high idle resources on Windows, Podman operates on a daemon-less and rootless-first architecture:
- Zero Idle Overhead: Since there is no background daemon running 24/7, Podman consumes virtually no memory or CPU when your containers are stopped.
- Fork-Exec Process Model: Every container runs as a direct child process of your shell session, avoiding single-point-of-failure engine crashes.
- Native Kubernetes Pods: Podman allows you to group multiple containers into a single shared namespace (a Pod), making local development match Kubernetes production environments.
- Docker-CLI Compatible: Podman matches the Docker CLI commands. You can simply alias
dockertopodmanand keep using your existing scripts.
Install
| Method | Command |
|---|---|
| Claude Desktop | just mcpb-pack → drag .mcpb onto Claude |
| Windows (NSIS) | just build-native → run *-setup.exe |
| Dev (any OS) | uv sync && .\start.ps1 → localhost:10806 |
Quick Start
git clone https://github.com/sandraschi/podman-mcp
cd podman-mcp
uv sync
.\start.ps1
Opens the web dashboard on http://127.0.0.1:10806 (API bridge runs on 10807).
What You Can Do
- "List all running containers and show resource usage."
- "Deploy my compose stack and verify every service is healthy."
- "Compare nginx:1.25 and nginx:1.26 — what changed?"
- "Analyze container my-app — why is it restarting?"
- "Back up my database volume before the upgrade."
Feature Overview
| Area | Highlights |
|---|---|
| Containers | CRUD, logs, stats, exec, inspect, health analysis |
| Images | List, pull, build, tag, push, prune, search, compare |
| Compose | Projects, up/down, logs, config, YAML file analysis |
| Backup/Restore | save/load image, backup/restore volume, export compose |
| Podman Machine | Status, hang detection, triple-kill recovery, restart |
| AI Chat | SSE streaming, tool execution cards, LLM provider discovery |
| Agentic | Deploy, cleanup, diagnose, rollback workflows |
| Prefab Cards | Containers, images, daemon status, system info |
Documentation
| Doc | Contents |
|---|---|
| docs/TOOLS.md | Full MCP tool reference |
| docs/COMPOSE.md | Compose management & file analysis |
| docs/BACKUP.md | Podman backup & restore guide |
| docs/CHAT.md | AI chat & agentic workflows |
| docs/ARCHITECTURE.md | Stack, transport, REST API |
| docs/CONFIGURATION.md | Environment variables |
| docs/DEVELOPMENT.md | Build, just recipes, testing |
| docs/TROUBLESHOOTING.md | Common issues |
Requirements
Python 3.12+, Podman Engine 20.10+, Node.js 20+ (dev). Rust 1.70+ (Tauri build only).
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。