container-manager-mcp¶
Manage Docker, Docker Swarm, and Podman containers through a typed MCP server and an integrated A2A agent — local or across a fleet of hosts over SSH.
Official documentation
This site is the canonical reference for container-manager-mcp, maintained
alongside every release.
Overview¶
container-manager-mcp exposes the Docker and Podman engines as a compact set of
action-routed MCP tools an agent or IDE can call, plus a Pydantic-AI graph agent
for conversational orchestration. It provides:
DockerManager/PodmanManager— typed Python facades over the Docker and Podman engines covering images, containers, volumes, networks, Swarm services, and Compose stacks.- Action-routed MCP tools — consolidated, togglable tool modules (info, image, container, volume, network, swarm, system, compose) that minimize LLM context overhead.
- Zero-script multi-host control — route any operation to a remote host over a
standard SSH channel, with no Docker TCP socket exposed, driven from a unified
inventory.yml(.yamllegacy fallback).
Explore the documentation¶
- Installation — pip, source, extras, and the prebuilt Docker image.
- Deployment — run the MCP and agent servers, Docker Compose, Caddy + Technitium.
- Usage — the MCP tools, the
DockerManagerPython API, and the CLI. - Overview — ecosystem role, enterprise readiness, and architecture.
- Multi-Host — zero-script Docker-over-SSH control plane.
- Concepts — the
CONCEPT:CMGR-*registry.
Quick start¶
pip install "container-manager-mcp[all]"
container-manager-mcp # stdio MCP server (default transport)
Run it as a network server:
See Installation and Deployment for the full matrix (PyPI extras, Docker image, all transports, the agent server, reverse proxy, and DNS).