The llms.txt File Builder generates the curated, AI-readable index that an increasing number of language models look for when they read your site. Raw HTML is noisy; an llms.txt file hands AI tools a clean Markdown map of your most important pages so they cite the right things. This tool walks you through the standard’s structure and emits a valid file ready to host.
How it works
The llms.txt format is plain Markdown with a fixed shape. The file opens with an H1 title (your site or product name), followed by an optional blockquote summary that states what you do in one line. After that come H2 sections — typically Docs, Guides, API, or About — each containing a bullet list of links. Each link uses Markdown syntax with an optional note after a colon, for example - [Quickstart](https://site.com/quickstart): Get running in five minutes. The builder takes your title, summary, and grouped links and renders exactly this structure, escaping nothing you do not need and dropping empty sections.
Tips and example
- Curate, do not dump. Link the 5–10 pages you most want an AI to read and cite, not every URL — the point is a high-signal map, not a sitemap.
- Write notes for ambiguous titles. The colon note after a link is your chance to tell the model what the page actually covers when the title alone is unclear.
- Serve it as text/plain at the root. Host it at
/llms.txt; consider an/llms-full.txtwith full page content for models that want deeper context.
Example output begins with # Acme Docs, then > Acme is an API for sending transactional email., then a ## Docs section listing your quickstart and reference pages as Markdown links.