pfsense-actions
A Docker app that gives scripts a few scoped pfSense actions and an operations console, without sharing a shell.
The problem
Home automation sometimes needs a small favor from the firewall, like checking that it's reachable, pinging a device, or waking a computer with Wake-on-LAN. The pfSense web interface isn't made for scripts, and handing a script SSH access gives it far more power than it needs.
What I built
pfSense Actions, a small self-hosted controller between your automation and one pfSense firewall. It offers four actions, a connectivity check, system info, ping and Wake-on-LAN, each behind its own scope. Every script gets a personal access token limited to the scopes it needs, and an operations console at /admin handles tokens, a redacted activity log, the pfSense connection, backups and alerts.
How it works
It's an Express service in TypeScript inside a Docker image supervised by s6-overlay. A request's token and scope are checked before anything touches the firewall, then the controller signs in over SSH, runs the fixed commands for that one action, and signs out. Callers get a status code and a small JSON result, never raw firewall output. The React console is protected by Argon2id passwords, an authenticator app with recovery codes, and passkeys, with all state in an encrypted SQLite database.
The outcome
A multi-architecture image that keeps the pfSense login in one place, lets you revoke any script's access at once, and alerts you through services like ntfy, Telegram and email when pfSense stops answering.
pfSense Actions began in 2024 as a little Docker controller with three endpoints, filter reload, Dynamic DNS refresh and Wake-on-LAN, all behind one shared API key. Over the next few releases it gained a ping check so a shortcut could tell whether the machine it just woke was actually up, plus rate limiting.
Version 2.0 in November 2026 narrowed it and grew it at the same time. The filter-reload and Dynamic DNS routes are gone, so the API covers connectivity, system info, reachability and Wake-on-LAN, and the single API key became per-script tokens with scopes. Around that sits a full operations console with first-run setup from a code in the container log, notifications, backup export and import, and debug logging, while the old routes still answer so existing scripts keep working.
The idea hasn’t changed since day one: give automation the pfSense actions it needs and nothing more.