Overview
The Docker MCP server provides an NLP-driven interface to manage Docker environments through an MCP protocol. It exposes resources to inspect and control containers, images, networks, and volumes, including per-container statistics and logs for debugging. The server supports natural-language prompts, notably the docker_compose prompt, which drives a plan+apply loop: describe the desired state, receive a concise plan, and apply or refine it iteratively. It supports resuming a project, exposing the current status of containers, volumes, and networks when a new chat begins. A WordPress deployment demo illustrates how natural language can drive real-world setups. Connectivity is via the Python Docker SDK’s from_env method, enabling operation against a local daemon (with docker.sock mounted) or a remote daemon reachable over SSH by configuring DOCKER_HOST. The server can be installed from PyPI (uv) or run in Docker, after cloning the repo. Safety considerations include avoiding sensitive data and excluding unsupported privileged modes. The README also describes development guidance and a developer-oriented workflow using Devbox for local testing.
Features
Natural-language container composition
Use natural language prompts to plan and deploy containers with a plan+apply loop.
Container introspection and debugging
Retrieve per-container CPU, memory stats and fetch logs for debugging.
Container lifecycle management
List, create, run, recreate, start, stop, and remove containers.
Image management
List, pull, push, build, and remove images.
Network management
List, create, and remove networks.
Volume management
List, create, and remove volumes.
Remote Docker connectivity (SSH)
Connect to a remote Docker daemon by configuring an SSH host URL.
Project state persistence and resume
Resume a project to restore the status of containers, volumes, and networks across chats.
Who Is This For?
- Server admins:connect to remote Docker engines for e.g. managing a public-facing website.
- Tinkerers:run containers locally and experiment with open-source apps supporting Docker.
- AI enthusiasts:push the limits of what an LLM is capable of.
