llms.txt
A plain-text file a website can publish to point AI systems at the content it considers most useful. It is a readable map, not a rule that is enforced, and it is an emerging convention rather than a settled standard.
llms.txt is a plain-text file that a website can place at its root to help AI systems find the content it considers most useful. The idea borrows the spirit of long-standing files like robots.txt: a simple, readable file at a known location that speaks to automated visitors. Where robots.txt says which parts of a site a crawler may request, llms.txt offers a curated map, pointing AI systems toward the pages and information a site most wants them to use and understand.
What it is
At its simplest, an llms.txt file is a short, human-readable list that highlights a site’s key content and links, written so that an AI system reading it gets a clear guide to what matters rather than having to infer it from the whole site. A site’s owner decides what to include. It is published by the site itself, which makes it a statement of what that site would like AI systems to focus on.
What it is not
Three clarifications keep expectations right.
It is not a standard. llms.txt is a proposed convention that some sites have adopted, not a formally ratified specification that AI systems are required to follow. Whether a given system reads it at all is up to that system.
It is not enforcement. Like robots.txt, it expresses a preference. It has no mechanism to compel any AI system to use it, ignore other pages, or behave in a particular way. It signals; it does not control.
It is not a guarantee of how content is used. Publishing one offers a map, and it does not determine whether or how an AI system quotes, cites, or represents the content it points to.
Who publishes one and where it fits
Adoption is uneven and still early. Sites that care about how AI systems read them, documentation sites and content-heavy publishers among them, are the most common adopters, while many sites have none. Because the convention is young, its usefulness will depend on how widely AI systems choose to honour it over time. For the bots the file is addressed to, see AI crawlers; for the broader practice it sits within, see GEO / AEO; and for where the file sits in a full AI-visibility architecture, Your CMS is a GEO decision.
We're booking content platform
engagements for 2026.
Twenty-five minutes to walk through the work and decide if we're the right team for it. Scoping and a fixed price come after.