Hi there,

we are currently testing a deployment of Anubis for Codeberg.org. However, we expect that the response code of error 200 results in a lot of unexpected breakage and support overhead, because debugging is harder.

For example, tools might correctly report connection issues if they receive an invalid response code. Serving data with "200 OK" could result in arbitrary issues like parsing issues, app crashes and much more. Even the Forgejo-native "webhook" test feature only checks for the 200 response code and would mark delivery as successful, although it might have been trapped by anubis.

I understand the motivation to trick crawlers into thinking they were successful, but this is probably not suitable for deployments that have legitimate and unknown automated traffic.