Free tool
Fill in what your product is and which pages are worth reading. The file builds as you type, in the markdown shape the convention expects. Copy it, drop it at the root of your domain, done.
llms.txt
# Your product name > One sentence on what this is and who it is for. ## Core pages - [Homepage](/): what the product is and who it is for - [Pricing](/pricing): plans and real numbers - [About](/about): who runs this
Everything runs in your browser. Nothing is sent to us or stored.
The temptation is to list every page you have. Resist it, because a file with sixty links is a sitemap with extra steps, and the thing that makes llms.txt useful is that a person chose what belongs in it.
Three habits make the difference. Write the annotation after each link as what the page answers rather than what it is called, since “pricing, with real numbers” tells an assistant more than “Pricing”. Use absolute URLs, because whatever fetches this may not know your origin. And say plainly what you are not for, which is the one thing marketing copy never does and the thing that stops an assistant recommending you to the wrong person.
Two things are worth checking once the file is live. First that it actually serves as plain text rather than your app shell, which is the failure that catches most single-page apps. Second that the crawlers can reach it at all, since an llms.txt behind a robots.txt block is a note nobody will read.
A plain markdown file served at /llms.txt that tells an AI assistant what your site is and which pages are worth reading. It was proposed by Jeremy Howard in 2024 as a way to give models a curated map of a site instead of leaving them to infer one from your navigation. The format is an H1 with your product name, a blockquote summary, then H2 sections of annotated links.
Honestly: unevenly, and no engine has committed to it the way search engines committed to robots.txt or sitemaps. Some assistants and dev tools fetch it, most do not announce whether they do. It takes about an hour to write and it cannot hurt you, which is the entire case for it. Treat anyone promising you traffic from it with suspicion.
At the root of your domain, so it resolves at https://yoursite.com/llms.txt, served as text/plain or text/markdown. A common mistake is a single-page app that returns the HTML shell for any unknown path, which makes /llms.txt look present while serving a web page. Load the URL in a private window and check you see plain text.
A companion convention that inlines the actual content rather than linking to it, so an assistant can read everything in one fetch. It suits documentation sites more than marketing sites. Start with llms.txt; add the full version only if your docs are the thing you want quoted.
robots.txt says what a crawler may fetch. A sitemap lists every URL you have. llms.txt is editorial: a short, curated set of the pages you would hand someone who asked what your product does. The value is in what you leave out.
No. The generator runs entirely in your browser. Nothing is posted to our API and nothing is stored.