How to generate an llms.txt file
- Enter the site name and a short summary. The name is the H1, which the llms.txt spec requires. The summary becomes the blockquote under it.
- Add the pages an agent should read. Group links under headings. Mark secondary pages as Optional. You can load URLs from the site sitemap, then delete the ones that do not belong.
- Copy or download llms.txt. Upload the file to the root of the site so it loads at /llms.txt. This generator does not publish it for you.
What the file is for
llms.txt is a proposed markdown file, documented at llmstxt.org, that gives an agent a short map of a site. The H1 is the site name. A blockquote can summarize it. Later sections are lists of links, each with a title and an optional note. Agents are expected to read the file and open the links that matter, not to treat the file as a crawl permission.
The spec's Optional section is the conventional place for secondary links. In the current version of the spec, that heading is a convention, not a switch that hides those links. This generator still puts secondary links under that heading so the file matches the published example.
What it does not do
It does not replace robots.txt. Crawl rules stay in robots.txt. If you want the links to match pages a crawler can find, build a sitemap with the XML sitemap generator, then spot-check one URL with the spider simulator. This page does not upload the file and it does not host a copy of your site.