$ ssh clawdbot.space --loading...
$ ssh clawdbot.space --loading...
A quick deployment walkthrough: a 10-minute VPS setup with security hardening, firewall rules, and an SSH tunnel that keeps your OpenClaw instance running 24/7 without exposing it to the public internet.
The walkthrough starts on a freshly provisioned Ubuntu VPS and immediately disables root login over SSH, creates a dedicated deploy user with sudo privileges, and enforces key-based authentication only. He treats every step as non-negotiable, reminding viewers that a default cloud image is a scanning target within minutes of booting.
The firewall configuration is deliberately minimal. UFW allows only the SSH port from a single trusted IP and the OpenClaw tunnel port on localhost. The video explains why it rejects exposing the agent's HTTP port directly: the agent should never be reachable by anyone who happens to know your server's address.
Instead of opening a port, the setup uses a reverse SSH tunnel so the agent's local interface is forwarded securely to a workstation. It wraps the tunnel in a systemd service with automatic restart, meaning even a reboot leaves the connection self-healing within seconds.
The final chapter covers persistence. OpenClaw itself runs under systemd as a managed service, with health checks that restart the process on crash and a log rotation policy that prevents disk bloat. A lightweight monitoring script is added that pings the agent every minute and sends a Telegram alert if it stops responding.
The payoff is a server you can forget about. The video closes on a 30-day uptime graph with zero manual interventions, making the case that a hardened VPS is the right home for an agent that needs to be available while you sleep.