llms.txt is a proposed Markdown file in the site root. It's meant to offer an overview of the important pages and short context for AI tools that choose to use it.
What is llms.txt?
The file usually lives at /llms.txt and contains the site name, a description and links to important resources. Its structure should be simple and readable for both people and machines.
It's no place for hidden claims. The information should match the public site and link to canonical pages where it can be verified.
Does it work today?
Adoption isn't universal, and the big platforms don't guarantee they'll use it. Google doesn't list it as a requirement for AI Overviews or AI Mode.
I'm testing it on this site as an experiment, together with tracking AI answers. The file existing proves nothing about impact, and it doesn't replace a good website.
llms.txt vs. robots.txt
robots.txt is an established protocol for crawling rules. llms.txt doesn't allow or block access. It only offers content for orientation.
Sensitive data doesn't belong in it. A public file shouldn't act as a security or access layer.