We tightened how parent domains are discovered during well-known file checks when your production URL uses a subdomain. This reduces the chance of sending requests to unrelated hosts while keeping the intended behavior of finding files on your main site.
Safer parent-domain probing
Parent-domain probing for files like robots.txt, sitemap.xml, and llms.txt now stops at the registrable domain, instead of relying on a simple label count. This prevents checks from walking into public suffixes such as co.uk, which are not owned by individual sites.
Checks also no longer attempt to derive “parent domains” from IP-based hosts. This avoids generating and probing unrelated IP-derived hostnames.