What is llms.txt
Also known as: llms.txt file, /llms.txt
llms.txt is a Markdown file, published at the site root, that summarizes the content and points AI agents to the most useful links, without the navigation, ads and code of a regular page. It is a proposal by Jeremy Howard, published in 2024. Google states that Search does not use these files: they neither help nor hurt on Google, but they may serve other systems.
ROO3's own site publishes an llms.txt at roo3.co/llms.txt. It starts with the company name as the title, follows with a quoted paragraph summarizing what the agency does and then organizes links into sections: services, projects, articles, this glossary, courses and contact.
Each link comes with a sentence saying what the agent will find there, including what the page does not promise. An agent that needs to answer questions about the company reads one short file instead of crawling dozens of pages full of menus and scripts.
The proposal's format
Version 2 of the proposal, from August 2026, also suggests offering Markdown versions of the pages themselves and linking them with alternate links.
- An H1 title with the site or project name, the only required part.
- A short summary in a blockquote.
- Paragraphs or lists with details, without new headings.
- Sections with H2 headings, each with a list of links in the [name](url) format, optionally followed by a note.
- A section called Optional, by convention, for links an agent can skip when it needs to save context.
What Google said in 2026
In June 2026, Google added a note on the subject to its AI optimization guide. Search, including its generative AI features, does not use llms.txt or other special AI files; having one neither harms nor helps visibility or rankings on Google, and keeping it for other services is fine.
The same guide lists llms.txt among the tactics you can ignore when the goal is Google Search. To appear in Google's AI features, the usual SEO practices apply.
llms.txt, robots.txt and sitemap
| What it is for | Who reads it | |
|---|---|---|
| robots.txt | Says what crawlers may crawl | Search and AI crawlers that follow the protocol |
| sitemap.xml | Lists pages for search engines to discover | Search engines, such as Google |
| llms.txt | Summarizes the site and points agents to useful links | AI agents and tools that adopt the proposal; not Google Search |
Who uses it, according to the proposal's author
In version 2, Jeremy Howard states that thousands of sites publish the file, that documentation platforms generate it automatically and that OpenAI, Anthropic and Google publish llms.txt files for their developer docs. The heaviest use, he says, is in software documentation, where coding agents follow the file to find technical references.
Is it worth having one?
For Google, it makes no difference. For other agents, it is a short file, cheap to maintain, that forces a company to write in one place what it does, what it does not do and where everything is. It makes sense as a complement, as long as nobody sells it as a shortcut to appearing on Google or in AI answers.
Frequently asked questions
What is llms.txt?
A Markdown file at the site root that summarizes the content and points AI agents to the most useful links.
Does llms.txt help Google SEO?
No. Google states that Search does not use llms.txt and that the file neither helps nor harms visibility or rankings.
Who created llms.txt?
Jeremy Howard, who published the proposal in September 2024 and a version 2 in August 2026.
What is the difference between llms.txt and robots.txt?
robots.txt says what crawlers may crawl. llms.txt neither allows nor blocks anything: it summarizes the site and points readers to links.
Do I need llms.txt to appear in ChatGPT?
There is no guarantee that a file will make an AI cite a site. It can help agents that read it, but what counts is content that is accessible, clear and trustworthy.
Related terms
- robots.txtRobots.txt is a text file at the root of a site that tells search engine crawlers which URLs they may or may...
- SitemapA sitemap is a file, almost always in XML, that lists a site's pages, videos and other files and provides...
- Structured dataStructured data is a block of code, invisible to visitors, that describes a page's content in a standard...
- IndexingIndexing is the stage where Google analyzes the content of a page it has already crawled, works out what it...
- SERPSERP stands for search engine results page, the page a search engine shows after a query. On Google, it...
- Content marketingContent marketing is a strategic approach focused on creating and distributing valuable, relevant and...
Rodrigo Fávaro
Founder of ROO3, a marketing and technology agency in São José do Rio Preto, Brazil. Builds AI products running in production (Tobia, gerar.app, Pense Mercado) and maintains the AI Benchmark, a public ranking of AI models.
X @rodmf LinkedIn rodrigofavaroMarketing you can measure, from the ad to the sale.
ROO3 builds and runs paid media, SEO and websites with real targets, dashboards and numbers. The first conversation is free.