llms.txt is a convention, not a standard, and no engine is known to require it.
We score it anyway, with no ranking claim. A file written for machines that asserts things your site does not is a defect, whoever reads it.
What it is
A plain Markdown file at /llms.txt that tells a machine
what your site is and links the pages worth reading, without parsing
your navigation. A longer companion, often llms-full.txt,
carries the text itself.
It is a proposal many sites adopted quickly. It is on no standards track, and we have found no engine that documents requiring it. Nobody can show you it is a ranking factor.
Why we score it
The file is a set of assertions about your site. If it names retired products, links pages that now 404, or describes a company that no longer matches your homepage, you have published a machine-readable inconsistency.
AC-02, machine-readable guidance quality looks for that. Serving no file is not scored as a failure. Serving one that contradicts you is.
Ours is generated from the same source as the sitemap and the pages, so the three cannot disagree. That is the part worth copying.
What a good one contains
- One sentence naming the organisation and what it does, matching the homepage.
- Links to the pages you want quoted, not every URL you have.
- Nothing that is not also true on the page it links, so every claim can be checked in one click.
- A generation step, so it changes when the site does.
You can read ours and the fuller one beside it.
Checking yours
The free scan fetches the file, reads it against the site it describes, and shows the captured response behind each finding. No account, no card, about twenty seconds.
The full rubric is published. This check sits in AI and agent controls.