Does Google use llms.txt?

The 30-second answer: does Google use llms.txt?

No, and Google says so in its own documentation. Google’s guide to optimizing for generative AI features, last updated 10 July 2026, states that you do not need to create machine-readable files, AI text files, markup or Markdown to appear in Google Search including its generative AI capabilities, because Google Search itself does not use them. The same section says that keeping such a file will neither harm nor help your visibility or rankings, because Google Search ignores it.

That covers classic Search, AI Overviews and AI Mode. Two other Google surfaces touch llms.txt without changing the answer: Google publishes one for its own Gemini API documentation, and Chrome’s Lighthouse tool has an optional audit for it. Neither is a Search signal.

So for Google, llms.txt is neither a lever nor a risk. Keep it for the agents and tools that do read it, and treat your Google visibility as an ordinary SEO problem, which is exactly what Google recommends.

What Google has said, in order

Google’s position has been consistent since the first public comment. The written guidance simply caught up with what its staff had been saying.

The June 2026 addition was reported at the time as a direct response to community questions, and its wording is deliberately permissive: creating and maintaining the file for other services that use it is described as completely fine. Google is not asking you to remove anything. It is telling you not to expect anything from it in Search.

AI Overviews and AI Mode: what actually governs inclusion

The generative features in Google Search draw on the same index as the ten blue links. Google’s documentation lists what you control, and none of it involves a new file:

• Being indexed. If a page is crawlable and indexable, it is eligible. noindex removes it from everything, AI features included.
• Snippet controls. nosnippet, data-nosnippet and max-snippet limit what Google may show from a page, which in turn limits what an AI Overview can quote.
• No special structured data. Google states there is no schema.org markup you need to add for AI features. Structured data still matters for rich results, which is a separate question.
• Google-Extended is not a Search control. That robots.txt token governs whether your content may be used to train future Gemini models and for grounding in Gemini apps and Vertex AI. Google’s crawler page says it does not impact inclusion in Search and is not a ranking signal.

In other words, the way to influence how you appear in AI Overviews is the way you influence how you appear in Search: clear pages that answer a question directly, with facts an assistant can quote.

The two places llms.txt does show up at Google

Google’s own developer docs

Fetched on 22 September 2026, ai.google.dev/gemini-api/docs/llms.txt is a valid llms.txt of roughly 200 links, each pointing at a Markdown version of a Gemini API documentation page. It exists for the same reason OpenAI and Anthropic publish theirs: coding agents and documentation tools read the format. That is a decision by a developer-documentation team about how to serve agents. It says nothing about Google Search, and Google’s Search documentation says the opposite.

Chrome’s Lighthouse audit

Lighthouse, the page-quality tool built into Chrome DevTools, gained a group of “agentic browsing” audits that includes an llms.txt check. Read the description carefully: it flags a page only if the server returns an error when the file is requested. If the file simply does not exist, the audit is marked not applicable, because, in the documentation’s words, providing the file is optional at the moment. It is a hygiene check for sites that chose to publish one, aimed at agents that browse, not a ranking factor.

Why Google takes this position

Mueller’s Reddit comment is the clearest statement of the reasoning, and it is worth taking seriously rather than dismissing. His points: no AI service had said it used the file; logs showed bots not requesting it; and, like the keywords meta tag of the early 2000s, the file is a site owner’s own claim about the site, which a search engine cannot trust without checking the pages anyway.

The fair counterpoint, and the one the spec itself makes, is that llms.txt was never designed as a ranking input. The v2 revision of the spec describes it as a guided path for agents that have already arrived at a site and need to find the right page without wasting context. That is a real use, but it is not search, and it is not what Google’s systems do. Both sides can be right at once: Google is correct that the file is not a Search signal, and the file can still be useful to the tools it was written for.

What to do if Google traffic is the goal

• Do ordinary technical SEO. Crawlable, indexable, fast pages. Check Search Console for coverage errors before anything else.
• Write quotable pages. A direct answer in the first paragraph, real numbers, a visible date. AI Overviews quote the same passages a featured snippet would.
• Use snippet controls deliberately. If there are pages you do not want summarised, nosnippet and max-snippet are the documented levers.
• Decide on Google-Extended separately. It is a training and grounding choice for Gemini, not a Search choice. Blocking it will not hide you from AI Overviews.
• Keep llms.txt, but budget zero Google effect. It costs nothing to maintain and other consumers read it. Just do not report it as a Google tactic.

What would change this answer

Google’s documentation is explicit, so the bar for revising this article is a change in that documentation: the mythbusting section being removed or reworded, or a Search Central announcement that Search or AI Mode consumes the file. A new Lighthouse audit or a new Google developer site publishing its own llms.txt would not qualify; those are about agents, not Search. Until then: Google does not use llms.txt, by its own account.

FAQ