Why it matters
- llms.txt is an optional Markdown file at the site's root that sums up for language models what the site is and where its main pages are, with links an agent follows instead of crawling the whole site.
- A file that is there with no title and no links does not do its job, and measuring tools such as Lighthouse count it as failed.
- Not having the file is no problem: it is a proposal not every provider uses.
How to fix
Write a title with the site's name, a line that describes it, then Markdown links to your important pages:
markdown
# Oud Store
> A Saudi shop for oud and incense, shipping across the Gulf.
## Pages
- [Products](https://example.com/products): every item and price
- [Shipping and returns](https://example.com/shipping): times and costsHow we detect
We ask for /llms.txt at the page's origin as ArablyzerBot, where robots.txt allows, with Accept: text/plain, and judge it as Lighthouse 13.6's llms-txt audit does: a 4xx means there is no file, which is no finding; a 5xx, a 429 or no answer is a finding; any other answer needs at least 50 characters, a # title line and one Markdown link. When the file answers and fails that, we ask for an address no site has: if the site answers that with 2xx too, it answers every address, and what we found is its catch-all answer, not an llms.txt, so there is no finding.