llms.txt generator
Your own words,
arranged for agents.
Write an llms.txt to the specification at llmstxt.org, or read one out of your site. Every line comes from a tag on your pages — nothing is written for you, and the file is checked against the specification as you edit it.
This tool works differently from the generator. The generator never uploads your image — it runs in your browser. The checker cannot: it fetches the public address you type, from our server, and reads your robots.txt for the sitemap it names, that sitemap, and then the HTML of up to twenty-five of the addresses in it — each page’s own title and description, and nothing else from any of them. Only the address you enter is sent. The addresses you check are then kept in this browser, so it can tell you what changed since last time; they go nowhere, and you can clear them.
Read it from your site
We take the sitemap your robots.txt names, then each page’s own <title> and its <meta name="description">. Nothing is composed: a page that declares neither is listed by address, and a gap stays a gap for you to fill.
Up to 25 pages, and the file says how many your sitemap declared. This fills the editor. Anything you have already written is kept.
The file
You have a file.
It is worth checking at the address it ends up at rather than the one it was written at. The checker reads what your server actually serves, against the same five rules.
Check the published fileQuestions
Does an AI write the descriptions?
No. Every line comes from a tag on your site: the H1 from your entry page’s title, the summary from its meta description, and each link’s name and notes from that page’s own title and description. A page that declares neither is listed by address with nothing beside it. A sentence a model wrote about what your site is for would be an inference in your voice, which is the one thing this will not produce.
How many pages does it read?
Twenty-five at most, and the page says how many your sitemap declared beside how many were read. The ceiling is real rather than a default: reading a site without one means one click can become an unbounded number of requests, and a number you cannot see is a truncation presented as a reading.
My sitemap is an index. What happens?
The first sitemap it names is read and the rest are not, and the page says so. Following every one of them is a walk through a document somebody else wrote with no bound on where it ends.
Does anything I type get sent anywhere?
No. The form composes the file in this tab and hands it back as a download; only the “read from your site” button reaches our server, and it sends the address you typed. Reading your site fills the form in rather than replacing it, so nothing you have written is lost by pressing it.
Where do I put the file once I have it?
At your site root, as llms.txt, or under the path it describes — the specification allows either and says a file covers the addresses beneath it. Then check the address with the checker: a file nobody can fetch is a file no agent reads.