What is llms.txt? With an example, a checker and a generator
llms.txt is a plain markdown file at the root of a website that tells AI systems what the site is and which pages matter, in a few hundred words. It is a proposal (llmstxt.org, 2024), not a standard any engine says it requires. Google says no such file is needed for its AI features. It is cheap, it does no harm, and it is a good place to state the facts you want repeated.
- A markdown file at /llms.txt: H1 with the site name, a blockquote summary, H2 sections of links. Proposed in September 2024.
- Google: not needed for AI Overviews or AI Mode. No chat engine documents it as an input.
- In our data the engines cite a business's own site in 0.6% of answers; the file cannot change what they read elsewhere.
- Worth having anyway: it costs ten minutes, states your facts in one place, and the checker and generator below do the work.
What is llms.txt?
We propose adding a /llms.txt markdown file to websites to provide LLM-friendly content.
llmstxt.org, The /llms.txt file
The idea is simple: HTML pages are built for browsers, with navigation, scripts and ads around the content, and a model reading them has to dig. An llms.txt is the site explained to a reader in a few hundred words of markdown, with links to the pages worth reading in full. The proposal is by Jeremy Howard, published 3 September 2024 and revised in August 2026, and it says thousands of sites now publish one, including the documentation sites of OpenAI, Anthropic and Google.
What goes in the file
| Part | Required | What it is |
|---|---|---|
| H1 | Yes, the only required line | The name of the site or project |
| Blockquote | Recommended | A short summary with the key facts needed to understand the rest |
| Paragraphs | Optional | Short notes: what the site offers, who it is for, where it operates |
| H2 sections | Optional | Each holds a list of links in the form [name](url): note, pointing to the pages that matter |
From the llmstxt.org specification, read 7 October 2026.
An example
This site's own file, trimmed, which you can read in full at citeon.ai/llms.txt:
# Citeon AI > Citeon AI is an AI search visibility platform. It measures what ChatGPT, Gemini, Claude, Perplexity and Google AI Overviews answer when buyers ask who to buy from, scores whether a business is named or recommended, and turns every gap into a fix grounded in the business's own website. ## Plans - [Pricing](https://www.citeon.ai/pricing): Visibility 59 euro a month, Optimize 199 euro a month, 3-day trial ## Pages - [Free AI visibility checker](https://www.citeon.ai/ai-visibility-checker): three engines, a report by email in 5 to 10 minutes - [Guides](https://www.citeon.ai/guides): GEO, AEO, AI Overviews, ChatGPT, llms.txt
Does llms.txt help?
You don't need to create new machine readable files, AI text files, or markup to appear in these features.
Google Search Central, AI features and your website
That is the only statement from an engine, and it says no. OpenAI publishes one requirement for ChatGPT search, that OAI-SearchBot can crawl the site; Perplexity and Anthropic publish none. So nobody promises to read the file. Our measurements add the other half of the answer: the engines cite a business's own site in 0.6% of answers and build the rest from Reddit, Wikipedia, review sites and directories. A file on your own site cannot change what they read elsewhere.
Why have one, then. It takes ten minutes. It puts your name, what you do, where you operate and your prices in one plain place that any crawler can read, which is the same discipline that makes your pages quotable. And it is harmless. Treat it as the last item on the list, after the questions are measured and the mentions are earned.
llms.txt checker
Enter a domain. The checker fetches /llms.txt from it and grades the format against the specification: found, plain text, H1, summary, sections, links, length. Nothing is stored.
llms.txt generator
Four fields, one file. It runs in your browser and sends nothing anywhere.
# Site name > One or two sentences on what the site is and who it is for. ## Key pages - [Homepage](https://example.com/): what the site offers
Save it as llms.txt at the root of your site, next to robots.txt. Then run the checker above.
llms.txt vs robots.txt
| Row | robots.txt | llms.txt |
|---|---|---|
| What it is | Rules for crawlers: what they may and may not fetch | A summary for readers: what the site is and where to look |
| Who reads it | Every well-behaved crawler, by convention and law in some places | Whoever chooses to; no engine documents it as an input |
| Effect of getting it wrong | Blocking OAI-SearchBot removes you from ChatGPT search answers | None documented |
| Where it lives | /robots.txt | /llms.txt |
If you do one thing today, check robots.txt: that OAI-SearchBot, PerplexityBot, ClaudeBot and Googlebot are allowed. Why that matters for ChatGPT.
Questions people ask
What is llms.txt?
A markdown text file at /llms.txt that gives AI systems a short, curated description of a site: an H1 with the site name, a blockquote summary, and sections of links to the pages that matter. Proposed by Jeremy Howard on 3 September 2024; the specification is at llmstxt.org.
Does Google use llms.txt?
Google's Search Central documentation says you do not need to create new machine readable files, AI text files, or markup to appear in AI Overviews and AI Mode. llms.txt is exactly such a file. For Google it is not needed; it is not harmful either.
Do ChatGPT, Claude or Perplexity read llms.txt?
None of them documents llms.txt as an input. OpenAI's published requirement for ChatGPT search is that OAI-SearchBot can crawl the site. The llmstxt.org proposal notes that OpenAI, Anthropic and Google publish llms.txt files for their own documentation, which is a signal of adoption by publishers, not a statement about what their search products read.
Is llms.txt the same as robots.txt?
No. robots.txt tells crawlers what they may fetch; llms.txt tells a reader what the site is about and where to look. One is permission, the other is a summary. Both sit at the site root.
What should an llms.txt contain?
The H1 with the site name (required), a blockquote summary, optional short paragraphs with facts an engine should know, then H2 sections holding link lists in the form [name](url): note. Keep it short and link to the pages rather than copying them.
How do I check my llms.txt?
Use the checker on this page: enter your domain and it fetches /llms.txt and grades the format. Or open https://yourdomain/llms.txt in a browser and compare it with the example.
Sources
- llmstxt.org, The /llms.txt file (Jeremy Howard, 3 September 2024; v2, 10 August 2026). the format: H1 (required), blockquote summary, H2 sections with [name](url): notes
- Google Search Central, AI features and your website. "You don't need to create new machine readable files, AI text files, or markup"
- OpenAI Help Center, Searching the web with ChatGPT. OAI-SearchBot access is the published requirement
- Citeon AI measurements, 24 August to 7 October 2026. own site cited in 0.6% of 7,552 answers