llms.txt
A proposed file at a site's root that points AI systems at its most useful documentation.
llms.txt is a convention, not a standard: a markdown file at the root offering a curated map of a site for language models, much as robots.txt offers rules to crawlers. It is cheap to publish, but measurement so far suggests answer engines read ordinary pages far more than they read the manifest.
Key facts
- llms.txt is a convention rather than a ratified standard, and no major engine has committed to honouring it.
- In one measured 15-day window, an answer-engine crawler fetched HTML pages 1,434 times against 4 fetches of llms.txt — roughly 360 to 1.
- A file served statically from the public directory bypasses application middleware entirely, so it is invisible to analytics unless routed through the app.
- Serving it costs almost nothing, which is why it is worth publishing even where the measured return is small.
Frequently asked questions
Is llms.txt worth publishing?
Yes, because it is nearly free — but do not count it as a traffic strategy. The measured ratio of page fetches to manifest fetches is heavily lopsided toward ordinary pages.
Why does my llms.txt never appear in analytics?
Because a static file in the public directory is served by the web server directly and never reaches your application. Route it through the app if you want it counted.
Machine-readable copy of this page:
/glossary/llms-txt.md