container-use

Why we included this project

If you have ever run more than a couple of coding agents on the same repo, you know how quickly they step on each other. One wipes out another's uncommitted work, or two agents fight over the same files. Container Use fixes that by putting each agent in its own container on a separate git branch, so several agents can work in parallel without conflicts and a failed experiment can be thrown away with zero cleanup. It is an MCP server that runs as a CLI tool with Claude Code, Cursor, and other MCP-compatible agents, using Dagger to handle the containerized workflows. The strongest part is visibility: you get the full command history and logs of what an agent actually executed rather than what it claims, and you can jump into any agent's terminal to take over when it gets stuck. For small teams that want to scale up agent usage without babysitting every task, this keeps agent work contained, reviewable through a normal git checkout, and under your control.