Release notes, 19 October 2026

Summary

If you used our Next.js steps for robots.txt, llms.txt or a file under /.well-known/, check that the file answers at its full address, such as /robots.txt. If it only answers without the extension, rename the folder to the full name, extension included. No check changed what it flags, so your scan results are the same.

Fixed

• Next.js: the builders now put a route handler in a folder named after the whole path, such as app/robots.txt/route.ts, as in Next.js's own example. We were dropping the extension, and app/robots/route.ts serves /robots, not /robots.txt.
• Guides: the example file is now labelled as an example whose values were chosen for our tests, and site-wide examples keep their example.com addresses instead of showing made-up paths on your domain. The robots.txt guide now points to the builders that write the file from your answers.
• robots.txt: the guide now says that a crawler named in its own group ignores the * group, because RFC 9309 sends a crawler to the * group only "if no matching group exists". Repeat any Disallow you want it to keep.
• llms.txt: the template no longer lists pages that may not exist on your site. Placeholders are marked as placeholders.
• nginx: the snippet now sets the content type itself rather than leaving it to the file extension.
• Content Signals: the docs now explain the optional use field (immediate, reference or full), which Cloudflare documents and our check already accepted.