A missing page no longer reveals where we store your files

Asking for a page that isn’t there used to get you a line of plain text naming the path we looked up inside the workload — something like file not found in workload: no-such-page/index.html.

That told anyone probing your site how our storage is laid out, and it distinguished a path that doesn’t exist from one they simply aren’t allowed to see. Neither is anything a stranger needs to know.

It also sent no cache headers. With nothing to go on, an intermediary cache is free to apply its own heuristics and hold on to a 404 as though it were the page that was missing — so a file that later appeared could stay missing for visitors long after you published it.

A request for a page that doesn’t exist now returns a plain, branded page with no internal detail in it, and an explicit cache policy of its own. The link back is relative, so it returns the visitor to your site rather than ours.

If you’d rather show your own design, you can — see your site’s own 404 page.