In short: llms.txt is a simple Markdown file in the root of a domain (e.g. example.com/llms.txt) that gives large language models (LLMs) a short, organised overview of the site's most important content. It is not an official standard and not a ranking factor: Google does not use it, while Anthropic (Claude) and Perplexity do. It is cheap to create, but it should not be the first item on your AI visibility to-do list.
What is llms.txt?
llms.txt is a Markdown-formatted text file that you place in the root directory of your website. Its purpose is to give AI systems a concise, structured overview of the site's most important content: key pages, documentation, policies. It was proposed along the lines of robots.txt and sitemap.xml, but the important difference is that it is not an official web standard: there is no IETF or W3C approval behind it, and not all of the major AI providers have confirmed that they actually use it.
The proposal was published by Jeremy Howard, co-founder of Answer.AI and fast.ai, on 3 September 2024, on the answer.ai blog and on llmstxt.org, which has since served as the reference. The problem it addresses: LLM context windows are finite, and converting a full website's HTML (navigation, ads, JavaScript) into LLM-friendly text is difficult and error-prone. This was a particular issue for developer documentation, where models need quick access to the relevant API descriptions while coding.
What is llms.txt good for, and what not?
Its practical role is narrower than many articles about it suggest. It is not a crawling instruction (like robots.txt), nor a complete table of contents (like sitemap.xml), but a context map. An AI system may use it when looking for an answer to a specific question about a given site, or when a developer tool loads documentation into its context.
Typical use cases:
- Developer documentation: API references and SDK descriptions in a quick, clean form. This is the most widespread and best-evidenced use.
- On-demand lookup: when a user of an AI search tool asks specifically about a brand or service, and the system checks whether there is an llms.txt.
- Retrieval workflows: some AI tools (e.g. Claude Desktop, Claude.ai) use the file in their own search logic when the user targets a specific domain.
What it is not good for: it does not affect Google's organic rankings, it does not replace robots.txt or sitemap.xml, and on its own it does not guarantee that an AI model will read it at all.
llms.txt, robots.txt and sitemap.xml – what is the difference?
Many people confuse the three, yet each serves a different purpose.
| File | Purpose | Audience | Can it block? | Official standard? |
|---|---|---|---|---|
| robots.txt | Access control: what the crawler may crawl | Search engine and AI bots | Yes | Yes (RFC 9309) |
| sitemap.xml | Full site map for fast discovery | Search engine crawlers | No | De facto industry standard |
| llms.txt | Curated summary of the most important content | LLMs and AI tools (partial, voluntary support) | No | No, a community proposal |
If you want to exclude AI crawlers (e.g. GPTBot, ClaudeBot, Google-Extended), you do that in robots.txt. llms.txt blocks nothing: it only speaks to systems that have already reached your site, and shows them what is worth looking at.
Which AI systems actually use it?
This is the most important question, and it is where most of the misinformation circulates. As of 2026:
- Anthropic (Claude): publicly confirmed. Claude Desktop and Claude.ai take the file into account in their search workflows.
- Perplexity: publicly confirmed. It fetches llms.txt and uses it to rank pages.
- OpenAI (ChatGPT): no official confirmation, only correlations observed by some teams.
- Google: a clear no. According to the Google Search Central documentation, Google Search does not use llms.txt, either for ranking or for AI Overviews. In July 2025 Gary Illyes confirmed that Google does not support it and has no plans to; John Mueller compared it to the ineffective keywords meta tag.
Some developer tools (code editors, documentation assistants) also read the file when the user explicitly provides a piece of documentation as a source.
How widespread is llms.txt?
According to an SE Ranking study covering 300,000 domains, adoption is only 10.13%: roughly one in ten sites has an llms.txt, after nearly a year and a half of industry talk. More importantly, independent analyses suggest that 97% of llms.txt files have never been requested by any AI system, and statistical models found no measurable citation effect between having the file and appearing in AI search.
We cannot fully confirm this, because AI providers' crawl logs are not public: the figures above are third-party estimates, not official statistics.
Is it worth creating an llms.txt file? – AdOps' view
The reality lies between two extremes. On the one hand, llms.txt is not a ranking factor, it does not guarantee an appearance in AI Overviews, and according to current data most of these files are simply never read by anyone. Anyone selling it as a miracle cure is misleading you. On the other hand, creating it takes minimal effort, it does no harm, and if support expands in future (as it already works with Anthropic and Perplexity), your site will be ready.
Our practical recommendation: do not put it at the top of your priority list. It has a far bigger impact if your content is structured, gives clear answers and is built on your own data – that is what AI search engines actually reward. In this process llms.txt is at most a cheap supplementary step, and it does not replace the content and technical fundamentals (see: SEO vs AEO vs GEO and AI Overview).
How is an llms.txt file structured?
The file follows Markdown syntax, because that is the "native language" of LLMs. The recommended structure:
- H1 heading with the name of the brand or site. This is the only required element.
- Blockquote summary: one or two sentences about what the site does.
- Context paragraphs: short, simple background information (optional).
- H2 sections: link collections grouped by topic.
- Link lists in the form
- [Page name](URL): short description.
A simplified example:
# AdOps Digital
> Digital marketing agency: SEO, AI optimisation, Google and Meta ads, social media.
## Services
- [AI optimisation](https://ad-ops.hu/en/ai-optimization): Visibility in ChatGPT, Gemini and Google AI Overview answers.
- [SEO](https://ad-ops.hu/en/seo): Technical, content and local SEO.
## Case studies
- [SEO case studies](https://ad-ops.hu/en/seo-case-studies/): Real client results with figures.
Alongside llms.txt you can optionally create an llms-full.txt, which combines the full text of the key pages into a single file so that an AI tool can read it in one request. This is mainly justified for large documentation sites (developer tools, API providers).
How to create an llms.txt file for a WordPress site
On WordPress you can create it in three ways: with a plugin, by manual upload or with a short piece of PHP code. For sites not built on WordPress, method 2 works: you simply upload the file to the root directory.
Method 1: plugin – automatic, maintenance-free
The simplest option is a dedicated plugin that automatically generates and updates the file based on your published posts and pages. Examples include the free Website LLMs.txt (wordpress.org) or, if you already use it, the built-in llms.txt generator in All in One SEO Pro.
- WordPress admin → Plugins → Add New.
- Search for "Website LLMs.txt", then install and activate it.
- The plugin makes the file available at
yourdomain.com/llms.txt. - Check: open the address in your browser.
Pros: it updates itself whenever new content is published, and no FTP is needed. Cons: irrelevant pages may also be included, and its content cannot be fully edited.
Method 2: manual upload – FTP or hosting file manager
If you want full control, create it by hand.
- Create a text file named
llms.txt(e.g. in Notepad or VS Code). - Fill it in following the Markdown structure above.
- Upload it to the root directory with an FTP client (e.g. FileZilla) or your hosting file manager (on WordPress, the folder containing
wp-config.php). - Check:
yourdomain.com/llms.txt.
Pros: full control over the content. Cons: it does not update itself; you have to extend it manually after every important new page.
Method 3: PHP code – a virtual file without a plugin
If you do not want yet another plugin, you can serve the file with a short piece of code. Put it in a wp-content/mu-plugins/llms-txt.php file or in your child theme's functions.php file:
add_action('init', function () {
if (strtok($_SERVER['REQUEST_URI'], '?') !== '/llms.txt') {
return;
}
header('Content-Type: text/plain; charset=utf-8');
echo "# AdOps Digital\n\n";
echo "> Digital marketing agency: SEO, AI optimisation, PPC management.\n\n";
echo "## Key pages\n\n";
echo "- [AI optimisation](https://ad-ops.hu/en/ai-optimization): Visibility in AI search engines.\n";
exit;
});
Pros: no FTP or extra plugin needed, and the code can be version-controlled. Cons: the page list has to be updated manually; keep the code in a child theme or an mu-plugin so that a theme update does not wipe it.
What to include, and what to leave out
Always include:
- your main service or category pages,
- your best and most visited expert content,
- the About us and Contact pages (this strengthens identification of the brand as an entity),
- the FAQ page, if it has its own URL.
Leave out:
- every single post – this is just noise and reduces the file's usefulness,
- login, thank-you and other system pages,
- noindex pages,
- duplicate URLs (tag pages, archives).
The value of llms.txt lies in being selective: a carefully compiled file of 10–30 entries is worth more than a 500-line, automatically generated list.
Checking and maintenance
- Open
yourdomain.com/llms.txt: the browser should display plain text. - Check the format against the specification on llmstxt.org.
- Check in Google Search Console that there are no crawl errors for the URL.
- When an important new page goes live, update the file (with a plugin this is automatic).
llms.txt is not a create-once file: if your site keeps growing, it is worth reviewing every six months.
Want to appear in AI answers more often?
llms.txt is only a small step. In our free AI visibility audit we show you whether you appear in ChatGPT and Gemini answers, and who they recommend instead of you.