llms.txt: What It Is, How to Write One, and What It Does Not Do
The most-searched file in AI SEO is also the least-read by AI. Here is the honest version: the spec, a worked example, how to add one on every common platform, and why you should spend twenty minutes on it and not a minute more.
llms.txt is a proposed standard for a markdown file at your site root that gives AI assistants a curated summary and link map of your site. It was proposed by Jeremy Howard in September 2024. As of 2026, no major AI search engine has confirmed using it for retrieval or ranking, so it is cheap to add but not a ranking lever.
That is the whole verdict. The rest of this guide exists because "cheap to add" is still true, and because the people searching for llms.txt deserve a straight answer about what it does before they are sold a generator.
What llms.txt Is
The proposal comes from Jeremy Howard, the fast.ai and Answer.AI co-founder, published on 3 September 2024 at llmstxt.org and revised as a v2 in August 2026. The idea is simple. A language model working with a website has a limited context window and cannot read everything. A sitemap lists every URL but says nothing about which ones matter. robots.txt says what may be fetched but not what is useful. llms.txt sits in between: a short, human-written markdown file that says what the site is, which pages matter most, and where to find them, in a format a model can read directly.
The proposal also suggests offering a markdown version of each page (the same URL with .md appended) and, optionally, an llms-full.txt that concatenates the whole documentation set into one file. Both of those came from developer documentation, which is where the idea has real traction: coding assistants fetching a library's docs to get an API call right.
Does Any AI Engine Actually Use It?
This is the question that matters, so here is what the vendors and the data say, not what generators claim.
- Google: John Mueller wrote in June 2025 that no AI system currently uses llms.txt, and compared it to the keywords meta tag. Gary Illyes said in July 2025 that Google does not support it and is not planning to. Google's Search Central page on AI features states that you do not need to create machine-readable files, AI text files or special markup to appear in AI Overviews or AI Mode.
- OpenAI, Perplexity, Anthropic: None documents llms.txt as an input to search retrieval or citation. Their crawler documentation describes robots.txt controls and nothing about llms.txt.
- The data: Ahrefs published a study in June 2026 covering 137,210 domains. About 28 percent had a valid llms.txt. Of those files, 97 percent received zero requests during May 2026 from anyone, bot or human. AI bots also never requested llms.txt on sites that did not have one, which is the clearest sign that they are not looking for it.
The proposal's own site notes that OpenAI, Anthropic and Google publish llms.txt files for their developer documentation. Publishing one is not the same as consuming one. The honest summary for September 2026: llms.txt has adoption among publishers and documentation tools, some use by coding agents on docs sites, and no confirmed role in how any major answer engine decides what to cite.
The Format
The spec is short. In order:
- An H1 with the site or project name. This is the only required element.
- A blockquote with a one-paragraph summary that gives a model the key facts.
- Optional paragraphs of extra context, with no headings.
- H2 sections, each containing a markdown list of links in the form [Page name](URL): one-line description.
- An H2 called Optional for secondary links an agent can skip when context is tight.
Plain markdown, no front matter, no JSON. If you can write a README you can write this.
A Worked Example for a SaaS Site
Here is a complete file for a fictional analytics product. Replace the names and it is done.
# Acme Analytics
> Acme Analytics is a product analytics platform for B2B SaaS teams. It tracks feature usage, funnels and retention, and connects to Segment, Stripe and HubSpot. Plans start at $99 per month with a 14-day free trial.
Acme is used by product and growth teams at companies from seed stage to Series C. The docs cover setup, the SDKs, and every report type.
## Product
- [Pricing](https://acme.example/pricing): Plans, limits and what is included at each tier.
- [Integrations](https://acme.example/integrations): Segment, Stripe, HubSpot and the data warehouse connectors.
- [Acme vs Mixpanel](https://acme.example/compare/mixpanel): Feature and pricing comparison.
## Docs
- [Quickstart](https://acme.example/docs/quickstart): Install the SDK and send the first event in ten minutes.
- [Events API](https://acme.example/docs/api/events): Endpoints, payload schema and rate limits.
- [Funnels](https://acme.example/docs/reports/funnels): How funnel reports are calculated.
## Company
- [About](https://acme.example/about): Founding team, location and mission.
- [Contact](https://acme.example/contact): Support and sales.
## Optional
- [Changelog](https://acme.example/changelog): Release notes by month.
- [Blog](https://acme.example/blog): Guides on product analytics and retention.Keep it under a couple of hundred lines. The point is curation; if you list every page you have rebuilt a sitemap in markdown. For a live example, Mavek's own file is at mavek.ai/llms.txt: an H1, a summary blockquote, a context paragraph, and H2 sections for core pages, agents, services, comparisons, integrations, blog, glossary and tools.
How to Add It, by Platform
Find your row, do it, and move on. Every method below was checked against the platform's own documentation in September 2026.
| You are on | Do this | Time |
|---|---|---|
| Next.js | Static: drop the file at public/llms.txt and it is served at the root. Dynamic: add app/llms.txt/route.ts with a GET handler that builds the markdown from your CMS or page list and returns it as text/plain. | 5 min |
| Webflow | Site settings > SEO tab > LLMs.txt section > Upload. Publish. Webflow serves it at yourdomain.com/llms.txt (not on the staging domain). | 2 min |
| WordPress (Yoast SEO) | Yoast SEO > Settings > Site features > llms.txt > toggle on. Yoast generates and updates the file for you; auto mode picks pages, manual mode lets you choose. Free since June 2025. | 1 min |
| Mintlify | Nothing to do. Mintlify hosts /llms.txt and /llms-full.txt for every docs site, plus a .md version of each page, and keeps them current from your page metadata. | 0 min |
| Astro, Hugo, plain HTML, anything static | Put llms.txt in the public or static folder of your project. It ships with the next deploy at the root URL. | 5 min |
| Shopify, Squarespace, other hosted | Most hosted builders have no native option yet. Check the app store or settings for an llms.txt feature; otherwise skip it. It is not worth a workaround. | n/a |
After deploying, open yourdomain.com/llms.txt in a browser and confirm it returns plain text with a 200 status, not an HTML 404 page dressed up as a file. That is the entire verification.
llms.txt Best Practices
If you are adding one, spend the twenty minutes well.
- Write the blockquote like a direct answer. What the product is, who it is for, what it connects to, what it costs. Those are the facts a model would want to quote, and the same facts should appear on your homepage and pricing page.
- Curate, do not dump. Ten to forty links across three to six sections. Pricing, comparisons, docs and about pages first.
- Use absolute URLs and keep descriptions to one line each.
- Keep it in sync. A stale llms.txt that lists retired plans is worse than none. Yoast and Mintlify regenerate theirs; on a static site, put it in the same review as your pricing page.
- Do not put anything in it that is not on the site. It is not a place for claims you would not publish elsewhere.
What llms.txt Does Not Do
It does not affect Google rankings, AI Overviews or AI Mode. Google has said so directly.
It does not get you cited in ChatGPT or Perplexity. Neither vendor documents it as an input, and the traffic data says their bots are not fetching it.
It does not control crawler access. That is robots.txt, and every major AI crawler honors robots.txt. If you want to allow search bots and block training bots, that is where you do it; our ChatGPT SEO guide has the crawler-by-crawler table.
It does not fix a page that is not indexed. If Bing has never seen your pricing page, ChatGPT search cannot cite it, llms.txt or not. IndexNow fixes that in three minutes.
And it does not substitute for a facts sheet. The consistency that makes a model trust your product description comes from your site, your docs and your third-party profiles agreeing, not from one markdown file.
What to Do Instead
The work that measurably affects whether AI engines cite you is unglamorous and none of it needs a new file.
- Crawler access: allow OAI-SearchBot, PerplexityBot, Claude-SearchBot and Googlebot in robots.txt. Decide about training bots separately.
- Indexing in both engines: Google via Search Console and clean indexing verdicts; Bing via Webmaster Tools and IndexNow.
- Answer-first pages with question-shaped headings and liftable facts. The full checklist is in our answer engine optimization guide, and the model-side view is in LLM SEO explained.
- Accurate schema where it matches visible content. Our schema generator covers Organization, Product and FAQ.
- A citation baseline: twenty buyer prompts, run monthly across ChatGPT, AI Mode and Perplexity, with the cited brands and URLs recorded. Mavek's AI visibility tracker automates that across ChatGPT, Claude, Gemini, Perplexity and Google AI Overviews, and our best AI visibility tools guide covers the alternatives.
Add the llms.txt if you like. Then go do the list above, because that is where the citations come from.
Frequently Asked Questions
What is llms.txt?
llms.txt is a proposed standard, published by Jeremy Howard in September 2024, for a markdown file at the root of a website that gives AI assistants a short, curated map of the site: a title, a one-paragraph summary, and sections of links with descriptions. It is meant to help a language model with a limited context window find the right pages quickly.
Does llms.txt help SEO?
No. Google has said it does not use llms.txt, and its documentation states that no machine-readable AI file is needed to appear in Search, AI Overviews or AI Mode. No other major answer engine has confirmed using it for retrieval or ranking either. It is harmless and cheap, but it is not a ranking or citation lever.
Do ChatGPT or Perplexity read llms.txt?
Neither OpenAI nor Perplexity has documented using llms.txt to decide what to retrieve or cite. Ahrefs' June 2026 study of 137,210 domains found that 97% of published llms.txt files received no requests at all in a month, and that AI bots never request the file on sites that do not have one. Treat any fetches as incidental, not as a signal.
What is llms-full.txt?
llms-full.txt is an optional companion file that contains the full text of your documentation in one markdown file, so an agent can load everything at once instead of following links. Documentation platforms such as Mintlify generate it automatically. It is only useful for developer docs, where a coding assistant may want the whole reference in context.
How do I create an llms.txt file?
Write a markdown file with an H1 (your site or product name), a blockquote with a one-paragraph summary, optional short paragraphs of context, then H2 sections containing link lists in the form [Page name](URL): one-line description. Save it as llms.txt at your site root. Yoast, Webflow and Mintlify can generate or host it for you.
Is llms.txt the same as robots.txt?
No. robots.txt tells crawlers what they may and may not fetch, and every major search and AI crawler honors it. llms.txt is a content guide with no permissions in it, and no major engine has confirmed reading it. If you want to control AI crawler access, robots.txt is the file that matters.