How to Install Docker MCP in Windsurf
Official Docker integration — manage containers, images, volumes and compose stacks via AI
What Docker MCP does
The official Docker MCP server gives AI agents full control over your Docker environment. Build images, run containers, manage networks, and orchestrate multi-service applications through natural language. Features: - List, start, stop, and remove containers - Build Docker images from Dockerfiles - Manage Docker volumes and networks - Run docker-compose up/down/scale - Stream container logs in real time - Inspect container health and resource usage - Push/pull images from Docker Hub and private registries - Execute commands inside running containers Use cases: - "Start the postgres and redis containers from my compose file" - "Show me logs from the api container in the last 10 minutes" - "Build a new image from the current directory and tag it v2.1" - "Which containers are using more than 500MB of memory?" Built and maintained by Docker. Connects to the local Docker socket — no remote credentials needed.
Installation steps
- 1Copy the config below
- 2Open ~/.codeium/windsurf/mcp_config.json
- 3Paste inside "mcpServers": {}
- 4Restart Windsurf
Configuration
File location: ~/.codeium/windsurf/mcp_config.json
{
"dockerMcp": {
"command": "npx",
"args": [
"-y",
"@docker/mcp-server"
],
"env": {
"DOCKER_HOST": "unix:///var/run/docker.sock"
}
}
}Install in other clients
For maintainers
If you maintain Docker MCP, add this badge to your README to show it's verified on CuratedMCP:
[](https://curatedmcp.com/marketplace/docker-mcp)
Related servers
GitHub MCP Server
Full GitHub API access for your AI agents — repos, issues, PRs, and more
Vercel MCP
Official Vercel integration — manage deployments, logs, env vars, and projects from your AI agent
GitLab MCP
Official GitLab integration — manage CI/CD pipelines, merge requests, and issues inside AI workflows