llms.txt generator

What is llms.txt?

A Markdown file at /llms.txt that gives AI agents a short map of your site. It is a proposal, not a standard, and nobody guarantees it is read. Sources checked on 2026-10-07.

llms.txt was proposed by Jeremy Howard on llmstxt.org in September 2024 and revised in August 2026 (changes). It is not an IETF or W3C standard. The idea: language models have small context windows and struggle with full HTML pages, so a site can offer a curated, plain-text list of its important pages.

The format

The file lives at the root of the site (https://example.com/llms.txt) and is plain Markdown, in this order:

  1. an H1 with the name of the site or project: the only required part;
  2. a blockquote with a short summary;
  3. optional free-text paragraphs or lists (no headings);
  4. H2 sections, each containing a list of links: - [name](url): optional notes;
  5. by convention, a section named Optional for links an agent can skip when its context is tight.

The 2026 revision also recommends offering Markdown versions of pages (page.md or page.html.md) and adds link relations so agents can discover them.

llms.txt example

# Atelier Grès

> Handmade stoneware tableware from Normandy. Small workshop, shipping in the EU.

## Main pages

- [Shop](https://shop.example.com/shop): All plates, bowls and mugs with prices.
- [Contact](https://shop.example.com/contact): Opening hours, address and phone.

## Blog

- [Choosing a mug](https://shop.example.com/blog/choosing-a-mug): Capacity, handle, glaze: our criteria.
- [Caring for stoneware](https://shop.example.com/blog/stoneware-care): Hot water, no abrasive products, air dry.

## Optional

- [Legal notice](https://shop.example.com/legal): Publisher, host, contact.

Real examples are easy to find: llmstxt.org publishes its own at llmstxt.org/llms.txt, and so does this site at /llms.txt.

llms-full.txt

llms-full.txt is the whole site (or documentation) concatenated into one Markdown file. It is not part of the llmstxt.org proposal; it is a convention some documentation sites adopted so a tool can load everything at once. It only makes sense for content that fits in a model's context.

Who actually reads llms.txt?

  • Google Search: no. Its AI optimization guide says you don't need "machine readable files, AI text files, markup, or Markdown" to appear in Google Search, because "Google Search itself doesn't use them", and that maintaining llms.txt "will neither harm nor help" your visibility.
  • Chrome Lighthouse has an llms.txt audit in its Agentic Browsing category. It calls the file "an emerging convention" and marks a missing file as "Not Applicable".
  • OpenAI, Anthropic, Perplexity: their crawler documentation is about robots.txt. None of these pages commits to reading llms.txt. OpenAI's bots page links to OpenAI's own llms.txt: that is publishing one, not reading yours.
  • llmstxt.org says coding agents "use them reliably". That is the proposer's statement, not an independent measurement.

In short: cheap to produce, possibly useful to agents and coding tools, no proven effect on rankings or AI citations. If you want a documented effect on AI crawlers, that is robots.txt: see robots.txt rules for AI crawlers.

How to create an llms.txt

  1. List the pages an agent should know about: home, product or service pages, documentation, key articles. Put legal pages under Optional.
  2. Give each link a short, specific note. A good meta description is usually enough.
  3. Save as UTF-8 plain text, publish at /llms.txt, and check that the URL answers 200 with a text content type (not your HTML app shell).
  4. Update it when important pages change.

Or let the llms.txt generator draft it from your sitemap in a few seconds, then edit it.

Limits worth knowing