Daemon unreachable¶

Symptoms: MCP tools return errors, repowire peer list is empty even when agents are running, the dashboard won't load.

Is the daemon running?¶

repowire status

Looks for the daemon on 127.0.0.1:8377. If it's down, start it:

repowire serve

That runs the daemon in the foreground. Useful for seeing errors directly. The installed user service (launchd on macOS, systemd on Linux) should auto-restart the daemon at login; if it doesn't, re-run repowire setup to reinstall the service.

Wrong port?¶

If you changed daemon.port in ~/.repowire/config.yaml, every hook still expects 8377 unless you've also reinstalled them. Either revert the port, or re-run repowire setup after the change.

Auth token mismatch¶

If daemon.auth_token is set in ~/.repowire/config.yaml, the daemon requires a Bearer token on WebSocket connections. Hooks and the MCP server read the same config and present the token, so this only breaks for external callers (your own scripts hitting the daemon). Confirm those callers send Authorization: Bearer <token>.

CORS or origin rejected¶

Browser-side errors that mention origin or CORS: the daemon restricts CORS to localhost origins (plus repowire.io when the relay is enabled). If you're hosting the dashboard somewhere else, that origin needs to be added explicitly — or front the daemon with a proxy you control.

Logs¶

repowire serve  # foreground, all logs to stderr

Or check the user-service log location reported by repowire status. Look for stack traces from FastAPI; routing errors are usually visible in the first 20 lines.

Process hanging on port 8377¶

Something else is bound to the port:

lsof -i :8377

Kill the offending process or change daemon.port (and re-run repowire setup).