robots.txt does not disallow llms.txt
Publishing llms.txt and then disallowing it in robots.txt is a contradiction that confuses every well-behaved agent. This check catches the rare but visible case where the two files are out of sync.
How the check decides
The check parses robots.txt and asks the universal user-agent * whether /llms.txt and /.well-known/llms.txt are allowed. Passes if both are reachable. Fails with the offending paths if either is disallowed. If no robots.txt exists the check passes (allow-all by default).
How to implement it
If you’ve added a global Disallow: / for any user-agent, add an explicit Allow: /llms.txt line above it for the same user-agent. Or just don’t disallow llms.txt in the first place.
Pass
User-agent: *
Allow: /
Fail
User-agent: *
Disallow: /llms.txt