docker-mcp
Enables interaction with Docker containers and images via tools and resources.
README
Docker MCP Server
This is a Model Context Protocol (MCP) server that provides tools and resources for interacting with Docker containers and images. It uses the stream HTTP transport method for communication and the Python Docker SDK (docker-py) for Docker operations.
Features
- List Docker containers and images
- Run, stop, and remove containers
- View container logs
- Pull Docker images
- Inspect container details
- Access Docker information as resources
Prerequisites
- Docker
- Docker Compose (optional, for easier deployment)
- uv (for local development)
Getting Started
Running with Docker Compose
- Clone this repository
- Build and start the server:
docker-compose up -d
Running with Docker directly
- Build the Docker image:
docker build -t mcp-docker-server .
- Run the container:
# Run server with SSE transport (no authentication)
docker run -d -p 3000:3000 -v /var/run/docker.sock:/var/run/docker.sock -e MCP_PORT=3000 --name mcp-docker-server mcp-docker-server
# Run server with SSE transport and token authentication
docker run -d -p 3000:3000 -v /var/run/docker.sock:/var/run/docker.sock -e MCP_PORT=3000 -e MCP_AUTH_TOKEN=your_secret_token --name mcp-docker-server mcp-docker-server
When the MCP_PORT environment variable is set, the server will use the SSE transport instead of stdio. Authentication is only applicable when using SSE transport. If the MCP_AUTH_TOKEN environment variable is set, the server will enable token authentication and configure the SSE endpoint to use the format /sse/{token} where {token} must match the value set in MCP_AUTH_TOKEN. If the token doesn't match or is not provided, the connection will be rejected. If no token is set, the server uses the standard /sse endpoint without authentication.
Running locally (development)
- Install uv (if not already installed):
curl -sSf https://astral.sh/uv/install.sh | sh
- Install dependencies using uv:
uv sync
- Run the server:
# Run with stdio transport (default)
uv run server.py
# Run with SSE transport on port 8000
MCP_PORT=8000 uv run server.py
# Run with SSE transport and token authentication
MCP_PORT=8000 MCP_AUTH_TOKEN=your_secret_token uv run server.py
Configuration
The server can be configured using environment variables:
MCP_PORT: When set, enables SSE transport mode and specifies the port to listen on (default:8000). If not set, the server will use stdio transport.MCP_AUTH_TOKEN: Secret token for authentication (only applicable when using SSE transport). When set, the SSE endpoint will be/sse/{token}instead of/sseMCP_HOSTNAME: Hostname to use for Docker containers (default:localhost). This can be a domain name or IP address that will be used to access Docker containers after they are created.MCP_AVAILABLE_PORTS: Comma-separated list of ports or port ranges that are available for Docker containers to bind to (e.g.,8000,8080-8090,9000). When specified, the server will validate that requested host ports are in this list.
Using with MCP clients
Should build the docker image first.
{
"mcpServers": {
"docker": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"-v",
"/var/run/docker.sock:/var/run/docker.sock",
"mcp-docker-server"
]
}
}
}
Available Tools
docker_ps: List running containersdocker_images: List available Docker imagesdocker_inspect: Get detailed information about a containerdocker_run: Run a Docker containerdocker_stop: Stop a running containerdocker_rm: Remove a containerdocker_logs: Fetch container logsdocker_pull: Pull a Docker image
Available Resources
docker://containers: Information about all containersdocker://images: Information about all imagesdocker://container/{container_id}: Detailed information about a specific containerdocker://config/hostname: Get the configured hostname for Docker containersdocker://config/available_ports: Get the list of available ports for Docker containers
TODO
- Add functionality to query documentation for a specific Docker image
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。