What is llms.txt?

How the audit reads it

The Discovery category is two checks on the files an engine reads before it reads you, and llms.txt is one of them. The audit fetches it from the site root alongside robots.txt and sitemap.xml. It first asks whether the file is really there: some servers answer a missing path with an HTML error page, so the audit checks that what came back is plain text and not an error in disguise. Then it asks whether the first line is a top-level heading, since that is what tells a parser it has found a document rather than a stray file.

The verdicts are graded. A missing file is a fail and scores 0. A file that exists but does not open with a heading is a warning and scores 0.5. Only two of the 21 checks can warn at all, and this is one of them; the other is AI crawler access.

What we do with a failed check

In Managed GEO, llms.txt is on the list of fixes our team ships, next to robots rules, headings, and page structure. We write the file, put it at the root, and re-run the audit so the row turns from fail to pass in the history. In GEO Consulting and Trends the check name goes on the prioritised list we hand your engineers, and they verify it in the same tool.

What it is not

It does not replace your pages and it does not carry your content. It is a short pointer that tells an assistant what the site is and which pages matter, so that the pages themselves, which have to pass the other checks, are the ones that get read.