llms.txt: what it is and how to write one

The idea was published in September 2024 by Jeremy Howard of Answer.AI, at llmstxt.org. The file is written in Markdown and lives at /llms.txt, next to robots.txt. Where robots.txt tells crawlers what they may fetch, llms.txt tells an AI tool what is worth reading.

1. One title line starting with #: the name of the site or business. This is the only required part.
2. A short summary in a quote block, a line starting with >.
3. Optional paragraphs with more detail.
4. Sections starting with ##, each a list of links with a short note.
5. A section named “Optional” for links that can be skipped when space is short.

# Example Bakery > A neighborhood bakery in Springfield. Open Tuesday to Sunday, 7am to 3pm. Closed Mondays. ## Facts- [Hours and location](https://example-bakery.example/visit): Opening hours, address and parking- [Menu and prices](https://example-bakery.example/menu): Current menu with prices ## Optional- [Our story](https://example-bakery.example/about)

Be realistic. llms.txt is a proposal. As far as we can tell, none of the large assistant companies has documented that its crawler reads the file, and Google has said it does not use it for Search. Some AI tools and developers do fetch it when they are pointed at a site. Treat it as a low-cost extra, not a fix.

If your facts are already clear in your pages and in structured data, llms.txt adds a little. If they are not, fix that first. The file is quick to write and harmless to publish, as long as you keep it true. An llms.txt with last year’s hours is one more stale source.

• Serving a web page instead of a text file at /llms.txt.
• Starting without a title line.
• Listing every page. Pick the few that answer real questions.
• Forgetting it when your facts change.
• Blocking it in robots.txt.