What is llms.txt, and does your site need one in 2026?

TL;DR

• llms.txt is a markdown file at the root of your site that hands AI agents a short summary and a curated list of links to your most useful pages.
• Google Search says it doesn't use llms.txt for rankings or for AI Overviews and AI Mode, and that keeping one for other services is fine.
• Chrome's Lighthouse checks for the file in its agentic browsing audits, and treats it as optional for now.
• Build one if agents read your product pages or docs. It takes about an hour. Don't expect traffic from it.

"llms.txt" gets around 33,100 searches a month worldwide, according to Google Keyword Planner. That is a lot of curiosity for a text file, and a fair amount of the advice ranking for it oversells what the file does. Here is what it is, what Google has said about it, and how we built ours.

What is llms.txt?

llms.txt is a proposed standard, first written up in 2024 and now on its second version at llmstxt.org. The proposal adds a markdown file at /llms.txt that gives language models brief background on a site, some guidance, and links to detailed markdown pages.

The reasoning is practical. Web pages are built for people, so the useful text sits wrapped in navigation, cookie banners, and JavaScript. An agent trying to answer a question about your product has to dig through all of that, and every wasted token costs time and money.

Think of your sitemap as the phone book. It lists every number in town. llms.txt is the friend who tells you which three to call first and what each one is good for.

How is llms.txt different from robots.txt and a sitemap?

They do separate jobs, and the proposal is clear about it. robots.txt tells automated tools which parts of a site they're allowed to access. llms.txt gets read on demand, when an agent needs information about a topic while it's helping someone.

A sitemap lists every indexable page, which on most sites adds up to far more than fits in a model's context window. It also rarely points to clean markdown versions of pages. llms.txt is curated by design, so the file stays small and the detail lives behind the links.

If you haven't sorted your crawler rules yet, start there. Our guide to robots.txt for AI covers which bots to allow and which to block, and it comes before llms.txt in priority.

What goes inside an llms.txt file?

The spec sets a fixed order. An H1 with the name of the site comes first, and it's the only required part. A blockquote with a short summary follows. After that you can add paragraphs or lists with more detail, as long as they contain no headings.

Then come sections under H2 headings, each holding a list of links. Every item is a markdown link, optionally followed by a colon and a note on what the page covers. A section called Optional, by convention, holds links an agent can skip when it's short on context.

The proposal also suggests offering a clean markdown version of each important page at the same URL with .md added. That way the links in your llms.txt point to text an agent can read without stripping out the page furniture.

What does Google say about llms.txt?

Google's guide to optimising for generative AI features, published on Search Central in May 2026, says you don't need machine-readable files, AI text files, markup, or Markdown to appear in Google Search, including its generative AI features. Search Engine Journal reported a later update that narrowed the wording to Google Search and added that it's fine to create and maintain llms.txt files for other services that use them.

So for Google rankings, AI Overviews, and AI Mode, llms.txt does nothing. The same guide puts its weight on unique, non-commodity content and crawlable pages instead.

Chrome is a different team with a different job. Its Lighthouse documentation lists llms.txt under agentic browsing audits and says the file helps agents avoid crawling a whole site to work out its structure. Lighthouse only flags a server error. If the file is missing, the audit shows as not applicable, because providing it is optional for now.

Which AI tools read llms.txt?

This is where the advice gets vague, so here is what we can show. The llms.txt proposal says the files are used most heavily for software documentation, where coding agents follow them to find API references. It also notes that OpenAI, Anthropic, and Google's Gemini publish llms.txt files for their own developer docs.

We saw this first hand while researching our post on MCP. Every documentation page our agent fetched from the MCP project and from Claude's docs opened with a note pointing agents to an llms.txt index. Documentation is where llms.txt has taken hold.

What we couldn't find is any AI vendor documenting that llms.txt affects whether a brand gets mentioned or cited in an answer. If someone sells you llms.txt as a way to rank in ChatGPT, ask them for the source.

Should a small business bother with llms.txt?

Yes, if you keep it in proportion. It costs an hour, it can't hurt your Google rankings, and agents now fetch pages while they answer questions for buyers. A file that states your pricing, who you serve, and what you integrate with is the cheapest way to hand those agents correct facts.

Do it last, though. Allow the search crawlers, keep your pages readable for agents, and publish content only you could write. Those three move your visibility. llms.txt tidies up the edges.

How did we build the llms.txt on contengi.com?

Ours sits at contengi.com/llms.txt and follows the spec. The H1 is the brand name. The blockquote says in two sentences what Contengi is and who it's for. Under that sits a key facts list covering who built it, what it runs on, pricing, where it publishes, and how to get in touch.

Next comes a short section on machine-readable versions. It points to llms-full.txt, which holds the full product documentation in one file, and explains that every page on the site is also available as markdown by adding .md to its URL. The rest is H2 sections for product, integrations, solutions, and guides, each a list of links with a one-line note.

We can't tie any traffic or AI mentions to the file, and we won't pretend otherwise. We track AI mentions separately with LLM tracking, and the file is one small input among many.

How do you create an llms.txt file in under an hour?

Start with the questions buyers ask before they buy, then pick the 10 to 20 pages that answer them. Pricing, the product overview, integrations, and your best two or three guides usually make the cut.

Write the H1 and the blockquote. Add a key facts list with the details an agent is most likely to get wrong about you. Group the links under H2 headings, give each one a short note, and put nice-to-have pages under Optional. The file should stay short enough that an agent reads it in one go, and every link in it need to earn its place.

If you run WordPress, the proposal lists Yoast SEO and AIOSEO as plugins that generate the file, and it says Wix creates one for every Wix site. Docs platforms like Mintlify and GitBook do it automatically too.

Then test it the way the proposal suggests. Give an agent only your llms.txt and ask it the questions a buyer would. If it gets your pricing or your audience wrong, the file needs work. Update it whenever pricing or products change, because a stale fact in a file built for machines spreads fast.

Frequently asked questions

Is llms.txt a ranking factor?

Not for Google Search. Google's generative AI guide says you don't need it to appear in Search or its AI features.

Where do I put llms.txt?

At the root, so yoursite.com/llms.txt. The proposal also allows a file at a subpath, such as /docs/llms.txt, covering the pages under that path.

What is llms-full.txt?

A companion file some sites publish, including ours, that holds the full content in one markdown file. It isn't part of the core spec.

Does llms.txt replace robots.txt?

No. robots.txt sets access rules for crawlers, and llms.txt is a reading guide an agent uses when it needs information.

Can WordPress generate llms.txt automatically?

Yes. The llms.txt proposal lists Yoast SEO and AIOSEO as WordPress plugins with generators.