Product loop
How OpenClowd works
OpenClowd connects an always-on machine, real repositories, terminal coding agents, and your browser so agent sessions can keep running on the host while you inspect and steer them from another device.
Last updated
1. Prepare an always-on host
Use a small VPS for the recommended path, or another reachable machine that stays online. Configure HTTPS and authentication; do not expose raw terminal, OpenClowd, or preview ports publicly.
2. Bring a real repository and your agents
OpenClowd assumes the repository exists on the host and the terminal coding agents you want to use are already installed and authenticated. It organizes those tools rather than replacing their setup and account flows.
3. Launch work inside a workspace
A workspace groups the repository, persistent terminal sessions, status, history, file and diff context, and local app previews. Multiple sessions can work in parallel without becoming anonymous terminal tabs.
4. Monitor, review, reattach, and steer
Use a browser on desktop or mobile to see what is running, spot attention requests, review changes, send short input, or reattach to the terminal session.