Subshell documentation
Subshell runs interactive coding agents in terminal sessions that you can access from any device.
Agents run on machines you own. Closing a browser tab doesn't stop the session, and opening the server from another device returns you to the same terminal.
What is Subshell?
Read What is Subshell? for an introduction to persistent sessions, remote execution, workspaces, and agent coordination, with examples from the interface.
Start using Subshell
Follow Get started to install Subshell Server, create your account, and launch a terminal. If you already have a server, open its address in your browser or follow Install Subshell Client.
- Installation covers desktop, CLI, Docker, and Proxmox deployment.
- Use Subshell covers sessions, workspaces, sharing, and notifications.
- Web previews & screens covers app previews, browser watching, and virtual display setup.
- Presets save agent configurations and launch defaults.
- Reusable Prompts save task instructions and ordered stacks.
- Manage nodes explains how to run agents on other machines.
- MCP and agent communication explains how agents exchange messages and coordinate work.
Run and extend Subshell
Administrators configure accounts, plugins, backups, and updates in Server administration. Use Networking to make the server reachable from your other devices.
For commands and configuration values, use Reference. For integrations, start with API and automation or Developers.
Product terms
A server is the control plane: the API, database, and web interface it serves. A node is a machine running agents. A client is your interface to a server.
The server's host can also run subshells. Enrolled nodes add other execution machines; they don't replace the server. Read Architecture for the component relationships.
Edit on GitHubLast updated on
