Site dependency scanner

One HTML page, as our server receives it: the addresses of its scripts, stylesheets, images, frames, media and forms, preconnect hints, and absolute URLs inside inline scripts, where fetch and WebSocket calls usually live. It follows up to 3 redirects, stops after 5 seconds or 2 MB, and never requests any of the hosts it finds.

The scanner reads the page as delivered, without running its JavaScript. Hosts that a script loads later, that only appear after sign-in, or that your server calls rather than the browser will not show up. You can still add monitors for them by hand.

When your payment provider, sign-in service or API backend is down, your site breaks even though your own server is fine. A monitor on each dependency tells you which one failed. The suggested monitors count any answer short of a server error as up, because many APIs reply 404 or 401 at their root address.