01
Paste a URL
We try to read the homepage for the title, meta description, and in-site links. If the site blocks cross-origin reads, you still get a starter file.
llms.txt is a short markdown file at the root of your site. It names the project, summarizes it, and lists the pages worth reading.
This generator drafts that file from your URL. It does not invent rankings. Major search engines have not said llms.txt is a ranking signal — it is still the cleanest way to point models at the right URLs.
The file follows the order in the llmstxt.org spec: one H1, a blockquote, notes, then H2 link lists.
We try to read the homepage for the title, meta description, and in-site links. If the site blocks cross-origin reads, you still get a starter file.
Link text from a nav is a hint, not a description. Rewrite each colon-note so an agent knows why the page exists.
Serve plain text or markdown from the site root. Add a link rel="describedby" in HTML if you want discovery hints.
As of 2026, Google has said you do not need llms.txt to appear in AI Overviews. Treat the file as documentation for agents, not as a guaranteed visibility boost.
Browsers cannot freely crawl other origins. The generator reads one homepage when a proxy allows it, then you curate. A host-side crawler is the right tool for large sites.