How to Create an llms.txt File for Your Website

To create an llms.txt file, write a plain markdown document named llms.txt, summarize your site and link to its key pages inside it, and publish the file at your domain’s root (for example, yourdomain.com/llms.txt). That’s the whole mechanical task. The harder question is whether it does anything: no major AI company has published documentation saying its crawler fetches or acts on an llms.txt file.

What an llms.txt file actually is

It’s a plain-text markdown file, proposed as a convention rather than a standard, meant to give an AI crawler a short summary of a site instead of making it parse full HTML pages. There is no enforced schema and no registry that checks whether you have one.

Whether anyone reads it

Google has published specific, narrow rules for the auxiliary files it does support. Its sitemap documentation says Google uses a sitemap’s lastmod field only “if it’s consistently and verifiably (for example by comparing to the last modification of the page) accurate,” and ignores hints like priority and change frequency (Google’s sitemap documentation). Its Indexing API documentation is narrower still: “The Indexing API can only be used to crawl pages with either JobPosting or BroadcastEvent embedded in a VideoObject” (Indexing API quickstart). Neither page mentions llms.txt. If Google or any other crawler reads the file today, it isn’t documented anywhere public.

That gap matters more if the goal is AI-assistant traffic specifically. Across seven websites we operate, AI assistants (ChatGPT, Perplexity, Gemini, Claude, Copilot) referred 26 of 3,842 sessions in a 90-day GA4 review — under 1% of total sessions. An llms.txt file might help a crawler parse a site faster if it bothers to fetch the file at all, but it isn’t going to move a number that small on its own.

How to write the file

Keep it to plain markdown. The common convention, used by sites that publish one, covers:

  • A one-line description of what the site or company does.
  • A list of links to the pages you most want summarized or cited, each with a short description.
  • Optionally, a link to a separate, more detailed llms-full.txt file if you maintain one.

A minimal example, line by line:

  • # Brand Name
  • > One sentence describing the site and who it's for.
  • ## Core Pages
  • - [Pricing](https://yourdomain.com/pricing): plan details.
  • - [Docs](https://yourdomain.com/docs): setup documentation.

How to publish it

  • Save the file as llms.txt — not .md or .html.
  • Upload it to your site’s public root directory, the same place as robots.txt.
  • Open https://yourdomain.com/llms.txt in a browser and confirm it loads as plain text, not a styled page.
  • If your site routes every request through a JavaScript framework, add a rewrite rule so requests to /llms.txt serve the static file instead of your app shell.

A bigger lever than llms.txt

Before spending time on a file no crawler has confirmed it reads, check whether your existing pages are indexed at all. On a September 2026 sweep of seven websites we operate, 419 of 1,482 inspected URLs — 28% — were not in Google’s index: some discovered but not crawled, some crawled but not indexed, some unknown to Google entirely. Fixing that is a bigger, more measurable lever than adding a new root-level file on the chance a crawler reads it.