llms.txt
Short definition
llms.txt is an emerging standard that lists a site's most important pages and summaries in a plain text file AI models can read easily.
Where robots.txt tells a crawler where it may not go, llms.txt works the other way round: it offers a site's most valuable pages, with short descriptions, in a plain list an AI model can read directly.
It is not yet an officially supported standard across every AI tool, but a growing number of sites add it as a readiness step. The file usually sits at the site's root, written in markdown, with a page title and a one-sentence summary for each entry.
llms.txt is often confused with sitemap.xml, but the two do different jobs. A sitemap lists every page on a site so search engines can crawl it exhaustively, while llms.txt is not a crawl list at all, it offers only a brand's handful of most valuable pages, each with a short summary, presented directly to a model.
Why it matters
When an AI model crawls a site and runs into complicated navigation or heavy JavaScript, it can misread the content entirely. llms.txt puts the most important pages in front of it in the simplest form the model understands.
Illustrative example
A digital agency wrote an llms.txt file listing its service pages with one-sentence summaries and placed it in the site root, then added updating the file to its publishing checklist for every new service page.
