A proposed root file offering a markdown map of a site's key pages for language models; put forward in 2024 but not adopted by Google or other major AI platforms.
llms.txt is a proposed file, placed at a site's root, that offers a curated markdown list of a site's key pages so a language model can read clean content instead of parsing full HTML. It was put forward in 2024 by analogy to robots.txt and sitemap.xml, as a way for publishers to hand AI crawlers a tidy map of what matters.
The proposal has not been adopted by the major AI platforms. Google has publicly dismissed it, with search staff comparing it to the long-dead keywords meta tag and noting that none of its AI systems use it; the other large model providers have not committed to reading it either. No major crawler is on record honouring the file, so on its own it changes nothing about how those systems retrieve or cite a site.
We serve one at /llms.txt anyway, alongside our Open Knowledge Format bundle, because it is cheap to generate from the same data and costs nothing if a future system does start reading it. Treat it as a low-effort bet, not as a working channel to AI visibility today.
