llms.txt is a Markdown file at the root of a site that tells language models what the site is and which pages are worth reading. It takes ten minutes to write. Below are a template to copy, what to list and what to leave out, and the limits: it grants no access, and Google says its AI features do not need it.
What is llms.txt for?
llms.txt is a proposal that Jeremy Howard of Answer.AI published in September 2024. A model that reads a normal web page has to dig the content out of navigation, scripts and layout. An llms.txt file hands it a short, plain list instead: the name of the site, one paragraph about it, and links to its most useful pages with a note on each.
A template to copy
# Example Bakery
> Example Bakery sells sourdough bread and baking classes in Leeds, with delivery inside the city and collection from one shop.
Prices are in pounds and include VAT. Classes run on Saturdays and must be booked online.
## Products
- [Sourdough loaves](https://www.example.com/bread): Six kinds of loaf, with ingredients, weights and prices
- [Baking classes](https://www.example.com/classes): Dates, what each class covers, and how to book
## Help
- [Delivery and collection](https://www.example.com/delivery): Areas, days, costs and cut-off times
- [Allergens](https://www.example.com/allergens): Every product with the allergens it contains
## Optional
- [Blog](https://www.example.com/blog): Recipes and news
What are the rules of the format?
The proposal defines these parts, in this order:
- An H1 with the name of the site or project. It is the only required part.
- A blockquote with a short summary.
- Paragraphs or lists of extra detail, with no headings.
- H2 sections, each holding a list of links written as
[name](url), with an optional note after a colon.
A section named Optional has a special meaning: its links can be skipped when a reader needs a shorter context. Put the long tail there.
What should the file list?
- The pages that answer what people ask about you: what you sell, what it costs, how it works, its limits.
- Pages that are true today. A stale link wastes the reader's budget.
- Pages that work without JavaScript or a login.
- Facts a model would otherwise guess, stated in the notes: prices, units, regions, what the product does not do.
Which mistakes are common?
Listing everything. The file is a curated list, not a sitemap. SEOAnything made this mistake itself: until 20 November 2026 its llms.txt listed 84 blog posts under the product pages. They now sit under "Optional", and the feature pages, free tools, comparisons and glossary come first.
Writing slogans. "The leading platform for growth" tells a model nothing. "Audits one URL and returns a score and a fix list" does.
Treating it as access control. llms.txt blocks and allows nothing. Access is the job of robots.txt, where each of the AI crawlers has its own user agent.
Linking pages a crawler cannot read. If the page needs JavaScript to show its text, the link leads to an empty document.
Forgetting it. Renamed pages and old prices stay in the file for months. Generate it from the same data as the pages, or review it when the site changes.
What is llms-full.txt?
It is a companion convention: one file holding the full text of the main pages as Markdown, for tools that prefer to read a single document. SEOAnything publishes one at /llms-full.txt, with every feature page, the comparison tables and the glossary definitions.
Does llms.txt help you get cited?
Be clear about what is known.
- Google's guidance on AI features says you do not need to create new machine-readable files or AI text files to appear in AI Overviews or AI Mode.
- The format is a proposal. No crawler is required to read it.
- It is useful today for coding assistants and agents that a person points at a site or its documentation, and it costs little to publish.
So write it, keep it short, and do not expect it to move rankings. The work that gets a page cited is the work described in GEO vs SEO: crawler access, text in the HTML, and passages that answer one question each. An XML sitemap is still what search engines read for discovery.
How do you publish and check it?
- Save the file as
llms.txtand serve it at the root:https://www.example.com/llms.txt. - Open the address in a browser. You should see the plain Markdown, not an HTML page.
- Open every link in it once. Each should answer 200; the free HTTP header checker shows the status code.
- Read the file as a stranger would: can you tell what the site does and where to look, without any other page?
The SEO tools produce a starter llms.txt, together with starter robots.txt and sitemap.xml files.
Sources: llmstxt.org, and Google's AI features and your website.
FAQ
Where does llms.txt go?
At the root of the site, so that it answers at /llms.txt, for example https://www.example.com/llms.txt. It is a plain Markdown file.
Is llms.txt required?
No. It is a proposal from September 2024, and no crawler is required to read it. Google says its AI features need no such file. It is useful for assistants and agents that a person points at a site.
What is the difference between llms.txt and llms-full.txt?
llms.txt is a short list of links with a note on each. llms-full.txt is a companion convention that holds the full text of the main pages in one Markdown file.
Comments 0
Join the conversation. Log in or create an account to like and comment.
Confirm your email to comment. The link is in your inbox.
No comments yet. Start the conversation.