MarkdownPaper Open reader

Guide

What is llms.txt?

llms.txt is a Markdown file at the root of a website, at /llms.txt, that tells AI agents and language models what the site is and links to its most useful pages. It is a short, curated map, written so a model can read it quickly instead of crawling every page.

You write

# Example Shop

> Example Shop sells refurbished cameras in the UK, with a one year
> warranty on every item.

Orders ship within two working days. We do not ship outside the UK.

## Help

- [Returns](https://example.com/returns.md): How to return an item within 30 days
- [Warranty](https://example.com/warranty.md): What the warranty covers

## Optional

- [Company history](https://example.com/about.md)

You see

Example Shop

Example Shop sells refurbished cameras in the UK, with a one year warranty on every item.

Orders ship within two working days. We do not ship outside the UK.

Help

  • Returns: How to return an item within 30 days
  • Warranty: What the warranty covers

Optional

Jeremy Howard proposed it in September 2024. The format is described at llmstxt.org, with a second version published in August 2026.

4 min read8 sectionsUpdated 1 Oct 2026

On this page

The format

The parts go in this order. Only the first is required.

  1. A # heading with the name of the site or project.
  2. A quote, starting with >, summing up what it is in a sentence or two.
  3. Any notes in plain paragraphs or lists, without headings.
  4. ## sections of links. Each item is a Markdown link, then optionally a colon and a short note on what the page holds.

A section named Optional is a common way to mark links a reader can skip when short of space.

Link to Markdown versions of pages

The proposal suggests offering a Markdown copy of each page, because it is far easier for a model to read than a page of HTML. Version 2 accepts either naming: page.html.md, with .md added to the address, or page.md, with the extension replaced.

A file for part of a site

A site can have more than one. A file at /docs/llms.txt covers the pages under /docs/, and the most specific file applies.

A real one

This site has one at markdownpaper.app/llms.txt. It starts with the name, a one paragraph summary in a quote, a paragraph about privacy and price, and then sections of links: using the reader, the desktop app, the guides and the comparisons. The build writes it from the same titles and descriptions as the pages, so it never falls out of date.

What it does and does not do

  • It helps agents that look for it. An AI agent sent to a site can read one file and know where to go next. Several AI companies publish one for their own documentation.
  • It is not robots.txt. It does not allow or block anything. To control crawling, use robots.txt.
  • It is not a ranking signal you can count on. Having one does not by itself make a site appear in search results or AI answers.
  • It is not a sitemap. A sitemap lists every page for search engines. llms.txt picks the pages worth reading and says what each one is.

How to write one

  1. Write the # name and a > summary of one or two sentences.
  2. Add the facts a model gets wrong about you, as short notes.
  3. Group your most useful pages under ## headings, each with a one line note.
  4. Save it as llms.txt at the root of the site, served as plain text.
  5. Keep it short. It is a guide to the site, not a copy of it.

Questions

What is an llms.txt file?

A Markdown file at /llms.txt on a website that summarizes the site and links to its most useful pages for AI agents and language models.

Is llms.txt the same as robots.txt?

No. robots.txt tells crawlers what they may visit. llms.txt describes the site and points to the pages worth reading. It allows or blocks nothing.

Does llms.txt help SEO?

Not directly. It can help AI agents that read it find the right pages, but it is not a ranking factor you can rely on.

Next