CoverM | Read coverage calculator for metagenomics | Genomics library

I believe the problem here is that you've configured your routes to match "נר-בינוני" unescaped, but the browser will actually percent-encode the URL before sending to the server, therefore the route matching actually operates on percent-escaped URLs. So the actual URL is https://coverme.co.il/product/%D7%A0%D7%A8-%D7%91%D7%99%D7%A0%D7%95%D7%A0%D7%99-tuberosejasmine/, and this does not match your route because %D7%A0%D7%A8-%D7%91%D7%99%D7%A0%D7%95%D7%A0%D7%99 is not considered to be the same as נר-בינוני.

EDIT: Unfortunately, using percent-encoding in your route pattern won't fix the problem, due to a known bug. Unfortunately, it's just not possible to match non-ASCII characters in a Workers route today. We intend to fix this, but it's hard because some sites are accidentally dependent on the broken behavior, so the fix would break them.

What you can potentially do instead is match against coverme.co.il/product/*, and then, inside your worker, check if the path also has נר-בינוני-tuberosejasmine. If it does not, then your fetch event handler should simply return without calling event.respondWith(). This will trigger "default handling" of the request, meaning it will pass through and be sent to your origin server like normal. (Note that you will still be billed for a Workers request, though.)

So, something like this: