- llms.txt is a proposal from llmstxt.org: a Markdown file, usually at
/llms.txt, that names your site, summarizes it, and lists links to the pages an AI tool should read first. - Only one part is required: an H1 with the site name. A blockquote summary, notes, and H2 sections of links are optional, plus an Optional section for links a tool can skip.
- Google says Google Search ignores llms.txt: creating one neither helps nor harms your visibility or rankings in Google Search, AI Overviews and AI Mode included.
- Chrome's Lighthouse has an experimental Agentic Browsing audit that looks for llms.txt. We found no official statement from OpenAI, Anthropic, Perplexity or Microsoft that their assistants read it.
- Write one if AI tools and agents use your docs or product pages; keep it short and accurate. Do not expect it to replace crawlable pages, robots.txt access or good content.
What is llms.txt?
llms.txt is a plain Markdown file, placed at /llms.txt on a website, that gives large language models and AI agents a short, curated overview of the site and links to its most useful pages. It was proposed by Jeremy Howard of Answer.AI on 3 September 2024, and the proposal lives at llmstxt.org, which was revised as a "v2" in August 2026.
The reasoning in the proposal is practical. Web pages wrap their content in navigation, scripts and ads, and an AI model's context window is limited, so reading a whole site is slow and expensive. A concise file written for machines gives an agent a cheaper starting point. The proposal says it is aimed mainly at use while answering a question (inference) rather than at training.
It is not a standard from a standards body, and it is not a crawling rule. It does not tell anyone what they may or may not fetch; that is the job of robots.txt, which our guide to allowing and blocking AI crawlers in robots.txt covers.
The llms.txt format
An llms.txt file is Markdown with a fixed order of sections, and only the first heading is required. The llmstxt.org proposal specifies:
llms.txt sections, in order
| Section | Required? | What goes in it |
|---|---|---|
H1 (# Name) | Yes, the only required part | The name of the site or project |
Blockquote (> …) | Optional | A short summary with the key information needed to understand the rest |
| Other Markdown, no headings | Optional | Paragraphs or lists with more detail: what the site is, what it is not, how to read the files |
H2 sections (## Section) | Optional, any number | Lists of links, each written - [Title](url): optional note |
## Optional | Optional, by convention | Secondary links an agent can skip when it needs a shorter context |
Three more details from the proposal:
- Location. The root (
/llms.txt) is the usual place, but the v2 proposal also allows a file at a subpath, covering the URLs beneath it. When several apply, an agent should use the most specific one. - Markdown copies of pages. The proposal suggests offering a clean Markdown version of useful pages at the same URL with
.mdadded (for examplepage.html.md), and linking them withrel="alternate" type="text/markdown". This part is separate from llms.txt itself and many sites skip it. - Tooling. The proposal lists generators and plugins, including documentation platforms (Mintlify, GitBook), WordPress SEO plugins (Yoast SEO, AIOSEO), Wix, and plugins for VitePress and Docusaurus. That is where most "llms.txt generator" options come from.
An llms.txt example you can copy
Here is a small, complete llms.txt for a fictional bakery website, following the llmstxt.org format. Replace every line with facts that are true for your site.
Example /llms.txt
# Example Bakery
> Example Bakery is a family bakery in Springfield that sells sourdough,
> pastries and custom cakes. We bake on site daily and deliver within the city.
We do not ship outside Springfield. Prices and opening hours on the linked
pages are the authoritative versions.
## Products
- [Bread and pastries](https://example.com/menu): the full daily menu with prices
- [Custom cakes](https://example.com/cakes): sizes, flavours and how far ahead to order
## Ordering
- [Order online](https://example.com/order): delivery areas, cut-off times and fees
- [Opening hours and address](https://example.com/visit): hours, parking and map
## Optional
- [Our story](https://example.com/about)
- [Blog](https://example.com/blog)Practical points, from our own experience of maintaining one (these are our practice, not part of the proposal):
- Say what you are not. A line about limits (no shipping, no free plan, not available in a country) stops tools from filling gaps with guesses.
- Keep it in sync. A file that contradicts your pages is worse than none. Generate the link list from the same data as your sitemap or navigation if you can.
- Return a real text file. Single-page apps often answer every URL with the HTML shell; check that
/llms.txtreturns status 200 with a text content type. - Do not put anything private in it. It is public, like robots.txt.
Does llms.txt help SEO?
For Google Search, no: Google says Google Search ignores llms.txt, so creating one will neither help nor harm your visibility or rankings, and that includes AI Overviews and AI Mode. Google's guide to optimizing for generative AI features lists "LLMS.txt files and other special markup" among the things you can ignore, saying Google Search itself doesn't use them. It adds that it is completely fine to maintain one for other services that do use it.
Google's separate AI features documentation says the same in general terms: you don't need new machine-readable files, AI text files or markup to appear in AI Overviews or AI Mode. The guide also notes that Google may crawl and index many kinds of files, but that being crawled does not mean a file is treated in a special way. If your llms.txt shows up in Google's index, that is not a sign it is being used for ranking.
What actually decides whether Google's AI features cite you is covered in our guide on how to get cited in AI Overviews: indexing, snippet eligibility and useful content.
Who reads llms.txt today?
Based on official statements we could open on 29 October 2026, Google Search does not use it, Chrome's Lighthouse checks for it in an experimental audit, and we found no AI assistant vendor that documents reading it.
Official positions on llms.txt
| Who | What they have published |
|---|---|
| Google Search (incl. AI Overviews and AI Mode) | Ignores llms.txt; neither helps nor harms rankings |
| Chrome Lighthouse | An experimental Agentic Browsing category includes an llms.txt audit that checks for the file at the domain root. A missing file (404) is marked not applicable because the file is "optional at the moment"; a server error fails the audit. |
| OpenAI, Anthropic, Perplexity, Microsoft | We found no official documentation saying ChatGPT, Claude, Perplexity or Copilot read llms.txt when answering. |
| Coding assistants and agents | Any tool can fetch the file if a user points it there or if its developer chose to. Your server logs show requests for /llms.txt, which proves a fetch, not use. |
The honest summary: llms.txt is most useful where a person or an agent goes looking for it, for example a developer pointing a coding assistant at your API docs, or an agent following the Lighthouse convention. It is not a channel that pushes your pages into chatbot answers.
How to write an llms.txt file in five steps
Writing one by hand takes under an hour for most small sites; a generator saves time only when you have hundreds of pages.
-
List the 10 to 30 pages that answer the most questions
Product or service pages, pricing, documentation, contact and policies. Leave out tag pages, thin posts and anything you would not want quoted.
-
Write the H1 and a two-sentence summary
Name the business and say what it does, for whom and where, in plain words. This is the part a tool is most likely to repeat.
-
Add a short note on limits
What you do not offer, and which pages are the authoritative source for prices or hours.
-
Group the links under H2s with one-line notes
Use
- [Title](absolute URL): what this page answers. Put secondary pages under## Optional. -
Publish and test
Upload it to the site root, open
https://yourdomain/llms.txtand confirm it loads as text with status 200. Re-check it whenever you change prices, plans or important URLs.
Check your llms.txt and AI crawler accessMrKeyword's Site Audit fetches /llms.txt on every crawl and reads your robots.txt rules for GPTBot, ClaudeBot, PerplexityBot and other AI crawlers.
Start freeHow MrKeyword uses llms.txt
MrKeyword publishes its own file at mrkeyword.io/llms.txt, and its Site Audit reports a missing llms.txt as a low-severity notice.
Our file follows the format above. It opens with # MrKeyword and a blockquote summary of what the product does and where its data comes from, followed by a paragraph on what it does not do (no comprehensive third-party backlink index, no competitor traffic estimates, AI visibility that reports stored answers rather than estimated AI traffic). One line says the file is a directory of public pages, not a ranking instruction or a guarantee of AI citations. Then come H2 link lists: Product, Developers, Guides, Guides and features, Free tools and Company. The "Guides and features" list is regenerated from the same page data as the site every time new pages are built, so it cannot drift from the pages themselves. We do not use an Optional section or publish Markdown copies of pages.
In the product, the Site Audit crawler requests /llms.txt once per crawl from the start URL's origin. It counts the file as present only when it returns a 2xx status with non-empty text that is not an HTML page, which catches the single-page-app problem above. If the file is missing, the audit adds a "No llms.txt" notice in its AI discoverability category, the lowest severity, alongside checks for AI crawlers blocked in robots.txt and content that needs JavaScript to appear. Separately, MrKeyword's AI page audit, which writes fix suggestions for individual pages, is instructed not to recommend llms.txt files, in line with Google's guidance that Google Search ignores them.
What MrKeyword does not do
MrKeyword does not generate llms.txt files, validate their Markdown structure, or measure whether any AI tool reads them.
- No generator. There is no llms.txt generator in the app; write the file by hand or use your CMS or docs platform's plugin.
- Presence only. The audit checks that a text file exists at
/llms.txt. It does not check for an H1, links or accuracy. - No usage tracking. MrKeyword cannot tell you whether ChatGPT, Claude or anyone else read your file. Its AI visibility feature records what those assistants answer to your questions, which is the outcome that matters.
Questions and answers
What is llms.txt questions
What is llms.txt?
llms.txt is a proposed Markdown file at the root of a website (/llms.txt) that gives large language models and AI agents a short overview of the site and links to its most useful pages. It was proposed by Jeremy Howard in September 2024 and is documented at llmstxt.org.
Does llms.txt help SEO?
Not in Google Search. Google's generative AI optimization guide says Google Search ignores llms.txt, and creating one will neither harm nor help a site's visibility or rankings, including in AI Overviews and AI Mode. It may help AI tools and agents that choose to read it.
Do ChatGPT, Claude or Perplexity read llms.txt?
We found no official documentation from OpenAI, Anthropic or Perplexity, as of 29 October 2026, saying their assistants read llms.txt when answering questions. Any tool can fetch the file if its developer or a user chooses to.
What should an llms.txt file contain?
An H1 with the site name is the only required part. Most files add a blockquote summary, a short note on details or limits, and H2 sections of Markdown links with a one-line note each, plus an Optional section for pages an agent can skip.
Is llms.txt the same as robots.txt?
No. robots.txt tells crawlers which URLs they may fetch. llms.txt is a reading guide that points AI tools at useful pages; it grants or blocks nothing.
Do I need an llms.txt generator?
Not for a small site: the file is a few dozen lines of Markdown you can write by hand. Generators built into documentation platforms and CMS plugins help when you have many pages that change often.
Sources
- The /llms.txt file — The llms.txt proposal (Jeremy Howard, Answer.AI), v2: format, location, Markdown pages and tooling
- Optimizing your website for generative AI features on Google Search — Google Search Central, on Google Search ignoring llms.txt and AI text files
- AI features and your website — Google Search Central, on not needing machine-readable files or AI text files for AI Overviews and AI Mode
- llms.txt audit (Lighthouse Agentic Browsing) — Chrome for Developers, on the experimental llms.txt audit
- Lighthouse agentic browsing scoring — Chrome for Developers, on the experimental Agentic Browsing category
- MrKeyword llms.txt — MrKeyword's own file, as described above
