Overview

The Docker MCP Server enables AI assistants to manage Docker directly through natural language. It provides 20 tools for controlling containers, images, networks, and volumes, plus a docker_compose prompt for deploying multi-container applications. The server connects to any Docker daemon — local or remote over SSH — making it suitable for development, server administration, and experimentation.

Community Server:

Developed by ckreiling with 710+ stars on GitHub. Python-based, installed via uv or pip.

Key Features

🚀

Container Management

Create, start, stop, remove, and recreate containers. Run containers with custom images and configurations.

🔍

Container Introspection

Fetch container logs, view CPU/memory stats, and inspect running containers in real-time.

🎻

Docker Compose

Deploy multi-container applications using natural language with the docker_compose plan+apply loop.

🔗

Remote Docker Support

Connect to remote Docker daemons over SSH for managing servers and cloud infrastructure.

🖼️

Image Operations

List, pull, push, build, and remove Docker images directly from AI conversations.

📀

Volume & Network Management

Create and manage Docker volumes for persistent data and networks for container communication.

Available Tools

Quick Reference

Tool	Purpose	Category
`list_containers`	List all containers	Containers
`create_container`	Create a new container	Containers
`run_container`	Create and start a container	Containers
`recreate_container`	Recreate an existing container	Containers
`start_container`	Start a stopped container	Containers
`stop_container`	Stop a running container	Containers
`remove_container`	Remove a container	Containers
`fetch_container_logs`	Get container logs	Containers
`list_images`	List available images	Images
`pull_image`	Pull an image from registry	Images
`push_image`	Push an image to registry	Images
`build_image`	Build an image from source	Images
`remove_image`	Remove an image	Images
`list_networks`	List Docker networks	Networks
`create_network`	Create a network	Networks
`remove_network`	Remove a network	Networks
`list_volumes`	List Docker volumes	Volumes
`create_volume`	Create a volume	Volumes
`remove_volume`	Remove a volume	Volumes

Detailed Usage

run_container▶

Create and start a container with the specified image and configuration.

use_mcp_tool({
  server_name: "mcp-server-docker",
  tool_name: "run_container",
  arguments: {
    image: "nginx:latest",
    name: "my-nginx",
    ports: { "80/tcp": 8080 },
    env: { NGINX_HOST: "localhost" }
  }
});

fetch_container_logs▶

Fetch logs from a running or stopped container.

use_mcp_tool({
  server_name: "mcp-server-docker",
  tool_name: "fetch_container_logs",
  arguments: {
    container: "my-nginx",
    tail: 50
  }
});

build_image▶

Build a Docker image from a Dockerfile or URL.

use_mcp_tool({
  server_name: "mcp-server-docker",
  tool_name: "build_image",
  arguments: {
    path: "/path/to/dockerfile",
    tag: "my-app:latest"
  }
});

create_volume▶

Create a Docker volume for persistent data storage.

use_mcp_tool({
  server_name: "mcp-server-docker",
  tool_name: "create_volume",
  arguments: {
    name: "my-data",
    driver: "local"
  }
});

Prompts

The server includes a docker_compose prompt that uses a plan+apply loop for deploying multi-container applications:

docker_compose: Use natural language to compose containers.
Provide a Project Name and a description of desired containers.
The LLM will present a plan for your approval before executing.

Example: "deploy a WordPress container and a supporting MySQL container, exposing WordPress on port 9000"

Installation

{
  "mcpServers": {
    "mcp-server-docker": {
      "command": "uvx",
      "args": [
        "mcp-server-docker"
      ]
    }
  }
}

Docker Socket:

Mounting /var/run/docker.sock gives the MCP server access to your local Docker daemon. For remote connections, set the DOCKER_HOST environment variable to an SSH URL (e.g. ssh://user@host).

Configuration

The server uses the Python Docker SDK's from_env method. Connect to a remote Docker daemon over SSH by setting the DOCKER_HOST environment variable:

{
  "mcpServers": {
    "mcp-server-docker": {
      "command": "uvx",
      "args": ["mcp-server-docker"],
      "env": {
        "DOCKER_HOST": "ssh://[email protected]"
      }
    }
  }
}

Security:

This MCP server has full control over Docker on the host machine. Do not configure containers with sensitive data like API keys or passwords. The server intentionally blocks dangerous options like --privileged and --cap-add/--cap-drop.

Common Use Cases

Development Environments: Spin up and tear down development containers with natural language
Server Administration: Deploy and manage containers on remote servers via SSH
Application Debugging: Fetch logs, inspect container stats, and debug running applications
Multi-Container Deployments: Use the compose prompt to deploy WordPress + MySQL or similar stacks
Image Management: Build, pull, push, and manage Docker images across registries

Docker MCP Server