Attent / Blog / Improving visibility

llms.txt: is it worth implementing?

llms.txt is a Markdown file describing a site with links for AI agents. How it differs from robots.txt, how to build it and why Google does not use it.

Illustration for the article: llms.txt: is it worth implementing?
In short

llms.txt is a plain text file in Markdown format, placed at /llms.txt, that briefly describes a site and points language models to the most important pages to read. It is not an access-rules file like robots.txt: it neither blocks nor allows anything. It is worth implementing if you have documentation, a knowledge base or a product that users ask AI assistants and agents about. Do not expect the file alone to improve your visibility in Google, though: Google Search does not use it.

What is llms.txt?

The llms.txt proposal was published by Jeremy Howard in September 2024 on llmstxt.org. In August 2026 a second version of the specification appeared, based on two years of experience in use.

The idea stems from a simple problem: a website is written for people. Menus, footers, scripts and ads take up space, and a language model with a limited context window has to pull the content out of all that. llms.txt gives it a shortcut: a short description of the site and a list of links to the pages that explain it best, ideally in a clean Markdown version.

The specification's author stresses that the file is meant mainly for the moment of answering a question (inference), not for training models. It is about the situation in which an AI assistant or agent visits a site to quickly understand a product and answer the user.

What llms.txt is not

The most common misunderstanding is treating llms.txt as a new version of robots.txt. They are two different files:

robots.txt llms.txt
Format directives for bots (User-agent, Disallow) Markdown: heading, description, lists of links
Purpose controlling bots' access to URLs pointing to the most important content and describing it concisely
Does it block anything yes, for bots that respect it no
Who reads it search engine bots and AI crawlers AI assistants and agents, developer tools

If you want to decide which AI bots can visit your site, you do that in robots.txt. llms.txt also does not replace a sitemap: it is a curated overview of the most important content, not a full index of URLs.

How an llms.txt file is built

The specification is short. The file lives in the root directory (/llms.txt) or in a subdirectory, for example /docs/llms.txt, in which case it applies to URLs under that path. It consists of:

  1. An H1 heading with the name of the site or project. This is the only required element.
  2. A quote (blockquote) with a one- or two-sentence summary of what the site is.
  3. Optional paragraphs or lists with important information that helps in understanding the rest.
  4. H2 sections with lists of links in the format [name](URL): short description.
  5. An "Optional" section with secondary materials that the agent can skip when it runs short of space in its context.

An example for a coffee shop might look like this:

# Sample Roastery

> Specialty coffee roaster from Kraków. We sell whole bean
> and ground coffee, and run a subscription and barista training.

- Shipping within 24 hours in Poland, free from $40.

## Offer

- [Whole bean coffees](https://example.pl/kawy.md): origin, flavor profile, roast level
- [Subscription](https://example.pl/subskrypcja.md): how it works, prices, cancellation

## Help

- [Shipping and returns](https://example.pl/wysylka.md): timing, costs, return procedure
- [How to brew](https://example.pl/parzenie.md): ratios and methods

## Optional

- [Roastery history](https://example.pl/o-nas.md)

The specification also recommends that important pages have a clean Markdown version at the same URL with .md appended (for example page.html.md). That way the agent does not have to process the whole HTML.

Who uses llms.txt

According to llmstxt.org, thousands of sites publish the file, and documentation platforms generate it automatically. OpenAI, Anthropic and Google (Gemini), among others, publish their own llms.txt files for developer documentation, and Lighthouse in Chrome checks for its presence as part of audits for browsing by agents. File generation is supported by Mintlify, GitBook, Wix, the Yoast SEO and AIOSEO plugins for WordPress, and plugins for VitePress and Docusaurus, among others.

Sites with technical documentation benefit the most today. Developers paste the contents of llms.txt into coding assistants, and agents working on a user's behalf reach for it to quickly understand a library or an API.

Does llms.txt improve visibility in Google and AI answers?

In Google, no. In its July 2026 guide, Google states plainly that to appear in search, including its generative features, you do not need to create special files, markup or Markdown versions, because Google Search does not use them. According to Google, an llms.txt file neither helps nor hurts visibility in search.

For other assistants the situation is less clear-cut. Large providers publish their own llms.txt files, but they do not state that their search, when answering user questions, uses other people's llms.txt files to choose which brands to recommend. Whether ChatGPT, Gemini or Perplexity mentions your company still depends first of all on what your site and external sources write about it: media, rankings, comparisons, reviews.

Treat llms.txt as a well-written business card for AI agents, not as a "get recommended" button. Implementing it takes an hour, but whether AI mentions your brand is decided by the content and sources the models read, and that has to be measured, not guessed.

Krzysztof CzapnikKrzysztof Czapnik, Chief Product Officer, Attent

When to implement llms.txt

The file makes sense when:

  • you have documentation, an API, a knowledge base or a help center used by developers and their assistants,
  • you sell a product with a complex offer (pricing, variants, terms) that users compare with the help of AI agents,
  • you want to give agents one place with a current, short description of the company and links to key pages.

It makes less sense on a simple brochure site where all the important information fits on a few clear subpages anyway. It will not hurt, but do not expect results from it.

How to implement llms.txt step by step

  1. Choose the 10–30 most important pages: offer, pricing, frequently asked questions, terms, documentation.
  2. Write a one- or two-sentence description of the company or product for the quote block: factual, without ad slogans.
  3. Group the links into H2 sections and add to each a short description of what the page covers.
  4. Prepare .md versions of the most important pages if your system allows it (some platforms do this automatically).
  5. Publish the file at /llms.txt and check that it returns plain text, not an error page.
  6. Update it along with your offer, because outdated prices or terms in a file for agents are worse than no file.

Summary

llms.txt is a simple, cheap-to-maintain file that organizes the most important information about a site for AI assistants and agents. It is worth having, especially with documentation and a broad offer. It is not, however, a way to gain visibility in Google or a guarantee that AI will start to recommend your brand. Whether ChatGPT, Gemini, Perplexity and Google AI mention you in answers, you can only find out through regular measurement, on the same questions, every week. That is what Attent does.

We check this for you, every week, in four engines

In a 30-minute call, we'll show you how AI answers your customers' questions today and who it recommends instead of you.

Book a call

Frequently asked questions

Is llms.txt the same as robots.txt?

No. robots.txt contains directives for bots and is used to block or allow URLs. llms.txt is a Markdown file with a short description of the site and a list of the most important links. It blocks nothing; it helps AI agents quickly understand what is on the site.

Does Google use llms.txt?

No. In its July 2026 guide, Google says that appearing in search and its generative features does not require special files or Markdown versions, and that llms.txt neither helps nor hurts visibility in Google Search.

Where should you put the llms.txt file?

In the root directory of the domain, at /llms.txt. The specification also allows files in subdirectories, for example /docs/llms.txt, in which case they apply to URLs under that path. The file should return plain text in Markdown format.

What must llms.txt contain?

The only required element is an H1 heading with the name of the site. Recommended are a short description in a quote block, important notes, H2 sections with lists of links and their descriptions, and an Optional section with secondary materials that the agent can skip.

Will llms.txt make ChatGPT start recommending my company?

There is no evidence for that. Whether an assistant mentions a brand depends mainly on the content on the site and on the external sources it reads: media, rankings, comparisons and reviews. llms.txt can make it easier for agents to understand your offer, but the effect has to be checked by measurement.

Sources

  1. llmstxt.org – The /llms.txt file, Jeremy Howard (2024, wersja 2: 2026)
  2. Google Search Central – Google's Guide to Optimizing for Generative AI Features on Google Search (2026)