Server Mode

XClaw isn't just a desktop app — its core can run as a standalone service, accessible in three ways.

The desktop client is the most complete experience, but it's not always available. You might be working on a remote server, checking progress on a tablet, or just wanting to run a quick task from the command line. Server mode makes these scenarios possible.

Once started, XClaw runs in the background, exposing a web interface and RPC API. The built-in browser UI provides core desktop client functionality — session management, model connections, tool usage — just without system-level integrations (notifications, shortcuts, etc.).

The command-line tool (CLI) offers yet another access method. Ideal for scripting and automation scenarios, or if you simply prefer working in the terminal.

All three access methods share the same backend. Sessions created in the desktop client are visible and continuable in the browser and CLI. Configuration and credentials are shared too.

How to

To start server mode, open a terminal and run the XClaw CLI command (see the installation docs for the exact command). Once the service starts, it outputs a local address (typically http://localhost plus a port number) — open that URL in a browser to access the web interface.

If you're using it on a remote server, be sure to configure access controls — by default, the service only listens on localhost. You can modify the listen address and port via the config file, or use an SSH tunnel for secure remote access.

The CLI approach is more direct — use commands to create sessions, send messages, and view results right in the terminal. Great for integrating into CI/CD scripts or other automation workflows.

Regardless of which access method you use, all data stays in sync. A task started on the desktop can be picked up in the browser.