container-use MCP
A Dagger MCP server that gives each agent its own container and git branch for safe parallel work
- 4,079 ★ on the repository
- Last commit
- Apache-2.0
Extended access
Why this level
- The agent runs arbitrary commands, but inside an isolated container
- Creates branches in the local git repository
Install
- Works in
Manual install
brew install dagger/tap/container-useFor all platforms there is a script: curl -fsSL https://raw.githubusercontent.com/dagger/container-use/main/install.sh | bash
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The agent runs arbitrary commands, but inside an isolated container
- Creates branches in the local git repository
This is third-party code. Review the repository files before installing.
What it does
container-use connects to an agent as an MCP server and moves its work into an isolated container with its own git branch. Several agents can work on different tasks at once without interfering with each other or your working copy. Full command history and logs are kept, and you can drop into any agent's terminal to take over. You review results with a regular git checkout of the branch and simply discard failed attempts.
Who it is for. For developers running several agents in parallel who want them isolated from their machine.
Good fit when
- You want several agents working on one repository at once
- You do not want the agent running commands directly on your machine
- You need to see what the agent actually did
Not a fit when
- Docker or a compatible container runtime is unavailable
- The task needs local devices or a GUI
Example request
Create a hello world app in Python with Flask in a separate environmentMCP
- Transport
- stdio
- Authentication
- not required
Limitations
The project is early and experimental and changes quickly. A container runtime is required, and work runs through the Dagger engine. Base images are pulled from public registries, and Docker Hub access from Russia can be restricted.
How to disable. Run claude mcp remove container-use or delete the server block from the MCP config, then brew uninstall container-use.
FAQ
How do I review an agent's work?
Switch to its branch with git checkout; all changes are there.
Is there a short command?
Yes, cu works the same as container-use.
Discussion
No questions or reviews yet.
Related
MCP serversA Docker container bundling browser, terminal, files, VS Code and MCP services for running agent actions safely in isolation
runpane MCP server: an agent in any repository opens Pane panels, starts other agents and reads their status
MCP Agent Mail
mcp_agent_mail
A mail coordination layer for coding agents: agents have memorable names, inboxes and outboxes, history search and voluntary file reservations
An MCP bridge: a single run_python tool runs code in a container and calls your MCP servers
A universal MCP CLI client from Apify: persistent sessions, OAuth, JSON output for scripts and shell access to any MCP server for agents
Shared memory for fleets of AI agents with access scopes, an audit trail and a knowledge graph: an MCP server, Docker and a cloud
Comparison with similar tools
Entries that share the main search query, and how they differ.
| Entry | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| container-use (this entry) | yes | not required | 4,079 | |
| Agent Safehouse | no | 2,125 | the safehouse command runs Claude Code | |
| VibeKit | no | 1,867 | vibeKit runs Claude Code | |
| Dory: Docker, Kubernetes and agent sandboxes on Mac | yes | not required | 1,633 | a Docker Desktop replacement for Apple Silicon Macs with a built-in |
README in short
The README describes isolated environments for agents and the main features: a container and branch per agent, command history, taking over. It gives install commands via Homebrew and a script, and a Claude Code setup example that appends rules to CLAUDE.md. Setup for other agents lives on the project site. Apache-2.0 licensed.