llms.txt generator
A short map of
your site, in Markdown.
An llms.txt generator turns your site’s name, a short summary and your key pages into a Markdown file in the format the llmstxt.org proposal describes, ready to save as /llms.txt.
Fill in the form, copy or download the file, and see whether it passes the three tests Lighthouse runs on it.
The honest status comes first: Google says its Search does not use the file, and neither OpenAI’s nor Perplexity’s crawler page says it reads one. Who does read it is below. It all runs in your browser and nothing you type is sent to us. Already have a file? Paste it under the form to check it.
By the Builderify team
A path that starts with / gets the site address put in front of it. A bare address works too, and its title is made from the last part of the path.
# Riverside Bakery > Family bakery selling sourdough, pastries and celebration cakes, for collection from the shop or by local delivery. Open Tuesday to Saturday, 7am to 4pm. Cakes need three days' notice. ## Menu and orders - [Bread and pastries](https://riversidebakery.example/menu/bread): Daily sourdough, rye and seeded loaves, with allergen notes - [Celebration cakes](https://riversidebakery.example/cakes): Sizes, flavours and prices - [How to order](https://riversidebakery.example/order): Collection times and the delivery area ## Visit us - [Find the shop](https://riversidebakery.example/visit): Address, opening hours and parking - [Contact](https://riversidebakery.example/contact): Phone and email ## Optional - [Our story](https://riversidebakery.example/about)
The three tests Lighthouse runs
- An H1 header: passes
- A Markdown link: passes
- 50 characters or more: passes
- i None of the links points at a Markdown version of a page (page.html.md or page.md). The proposal suggests that where you have one, and plain page links still work; they just leave the navigation and scripts for the reader to cut away.
- i The Optional section is kept. Version 2 of the proposal still allows it as a convention for secondary links, but it no longer carries any special meaning for tools.
Already have an llms.txt? Check it and load it here
Your pasted file, as it was
How an llms.txt file
is put together.
It is an ordinary Markdown file with a fixed order, so that a program can read it as easily as a person. One part is required; the rest are there to make the file useful.
| Part | Written as | Required | What the proposal says |
|---|---|---|---|
| The name | # Riverside Bakery | Yes | An H1 with the name of the site or project. The proposal calls it the only required section. |
| A summary | > Family bakery selling sourdough… | No | A blockquote with the key information needed to understand the rest of the file. |
| Details | Open Tuesday to Saturday… | No | Paragraphs, lists or any other Markdown, except headings. |
| File lists | ## Menu and orders | No | H2 sections, each a Markdown list. Every item has a required link, then optionally a colon and a note: - [Title](https://…): note |
| Optional | ## Optional | No | By convention, secondary links. Version 2 still allows it, but it no longer carries special meaning for tools. |
From the llmstxt.org proposal. Version 2 (August 2026) also lets a file sit at any path and cover the pages under it, with the most specific file applying(see what changed in version 2).
What anyone has said
about reading it.
Making the file is cheap. Whether it earns you anything depends on who reads it, and the public record is thin. This is what each source says, opened on 20 November 2026.
| Reader | What its own pages say | Source |
|---|---|---|
| Google Search | Says it does not use the file. Its guide says Google Search "ignores them", and that keeping one will "neither harm nor help" visibility or rankings in Search. | Google, updated 10 Jul 2026 |
| Lighthouse | Fetches /llms.txt at the root of the site as part of its Agentic Browsing category, which it calls experimental. A 404 is marked not applicable because the file is optional "at the moment"; a server error fails it. | Chrome for Developers |
| OpenAI | The crawler page lists OAI-SearchBot, GPTBot and ChatGPT-User and what robots.txt does for each. It mentions llms.txt once, as a pointer to OpenAI's own documentation index, and says nothing about reading other sites' files. | OpenAI, opened 20 Nov 2026 |
| Perplexity | The crawler page lists PerplexityBot and Perplexity-User and their robots.txt tags. Like OpenAI's, it mentions llms.txt only as the index of its own documentation. | Perplexity, opened 20 Nov 2026 |
| Coding agents | The proposal's author says the files are used most heavily for software documentation, where coding agents follow them to API references and tutorials. That is his statement, not a count. | llmstxt.org |
What server logs show
EZY Research watched 83 sites that all published an llms.txt, from 27 April to 19 July 2026, and counted how often each crawler asked for it next to how often it asked for robots.txt(their study). The AI crawlers that read robots.txt thousands of times asked for llms.txt a handful of times.
Treat it as one data point. EZY sells an AI-visibility tool, its panel skews towards small-business sites, and it says so itself. A fetch also shows that a file was requested, not that it changed an answer.
| Crawler | robots.txt | llms.txt |
|---|---|---|
| OpenAI | 3,990 | 7 |
| Anthropic (ClaudeBot) | 3,120 | 9 |
| PerplexityBot | 775 | 0 |
| Googlebot | 5,125 | 67 |
| Meta-ExternalAgent | 172 | 193 |
Fetches over 12 weeks, as published by EZY Research. Meta’s crawler is the one that asked for llms.txt more often than robots.txt.
Our reading: an llms.txt file is worth ten minutes if you want a tidy map for the tools and coding agents that do use one, and for Lighthouse’s check. It is not a way to rank. The guide to getting your website into ChatGPT covers what is more likely to matter.
Where the file
has to live.
At the root of your domain, so that it opens at https://yourbrand.com/llms.txt. That is the address Lighthouse requests, and the one the proposal names first.
This site publishes one at /llms.txt, generated from the same page list as its sitemap, so you can see a long one in the wild.
Use the Copy or Download button above, and name the file exactly llms.txt.
It has to open at /llms.txt on your domain. If your platform cannot serve a plain file at the root, the proposal also allows one at a sub path such as /docs/llms.txt, which covers the pages under that path.
You should see the text. A missing file answers 404, which Lighthouse marks not applicable; an error from the server fails its audit, so a broken file is worse than none.
Version 2 suggests rel="describedby" to point from a page to the llms.txt that covers it, as a link element in the head or as an HTTP header: <link rel="describedby" href="/llms.txt"> or Link: </llms.txt>; rel="describedby".
The proposal lists Yoast SEO and All in One SEO for WordPress, and Wix, as generating an llms.txt file. Paste what yours produced into the checker above to see what it says.
From the proposal, its changes page and Lighthouse's documentation, read on 20 November 2026.
A curated list,
not a copy of your sitemap.
The proposal says a sitemap lists everything while llms.txt offers a curated overview, and that the file itself should stay small enough to fit in a model’s context.
The proposal’s guidelines
- Use concise, clear language.
- When linking to resources, include brief, informative descriptions.
- Avoid ambiguous terms or unexplained jargon.
- Test your file by asking an agent questions about your content, giving it only your llms.txt as a starting point.
Our advice, not the proposal’s
- List the pages you would send a new customer to: what you sell, what it costs, how to reach you, and your returns or booking rules.
- Write each note as the answer to the question that page settles: “Prices and times”, not “Learn more”.
- Put anything you would skip in a hurry under Optional.
- Edit the file when a page moves or goes. A list of dead links is worse than no list.
Guidelines quoted from the llmstxt.org proposal, opened on 20 November 2026.
Seven ways a file
goes wrong.
The generator fixes what it can and says so in the list under the file. The same rules sit behind the checker for a file you paste in.
| Mistake | The rule | From |
|---|---|---|
| No H1, or not first | The H1 is the one required part and it comes first. Lighthouse fails a file with no line that starts with # and a space. | Proposal Lighthouse code |
| Headings in the details | Between the summary and the first H2 the proposal allows any Markdown except headings. The generator removes the # marks and keeps the text. | Proposal |
| A list that is not links | Each file-list item needs a Markdown link, [name](url). A note can follow after a colon. Lighthouse fails a file with no link at all. | Proposal Lighthouse code |
| A path with no domain | Every example in the proposal uses a full address. A reader that picks the file up away from your site has no domain to add, so the generator puts yours in front of any path that starts with /. | Proposal |
| Brackets or parentheses in a title or address | They end a Markdown link early. The generator removes brackets from titles and writes ( and ) in addresses as %28 and %29. | Markdown itself |
| An error from /llms.txt | Lighthouse flags a response of 500 or above, and 429. A plain 404 is not applicable, so having no file is safer than a broken one. | Lighthouse docs Lighthouse code |
| A file under 50 characters | Lighthouse calls it suspiciously short. A name and one link will clear it. | Lighthouse code |
The three tests are the ones in Lighthouse's source: a line that starts with a hash and a space, a Markdown link, and at least 50 characters. Lighthouse’s category is experimental and needs Chrome 150 or later. Passing them says the file is well formed, not that anything reads it.
If you want a bot that
knows your pages.
An llms.txt file is a map for readers you do not control. If what you want is for your own visitors to get answers from your pages, you can teach a bot those pages directly, and no outside crawler has to read anything.
Paste text
The module page says to drop in a policy, a spec sheet or an about page: anything you would otherwise have emailed somebody.
Give it a URL
It fetches a page you already publish and reads it; the same URL is only ever stored once. The pages in your llms.txt are a ready list of the ones to start with.
Write a Q&A
For the question you get every week, write the question and the exact answer you want given back.
The AI Chatbot answers from the knowledge you give it and hands over what it cannot answer; how to train a chatbot on your website content covers what to add. Builderify is free to start, with 250 free credits a month and no monthly fee: see pricing.
Start freellms.txt, before you ask.
It is a Markdown file at /llms.txt that gives language models and agents a short, curated overview of a website. The proposal comes from Jeremy Howard (September 2024; version 2 is dated 10 August 2026). The only required part is an H1 with the site's name. After it can come a blockquote summary, some details, and H2 sections that list links with an optional note on each.
Type your site name, a one or two sentence summary and the pages a newcomer needs, one a line as Title | address | note. Copy or download the result, save it as llms.txt and upload it to the root of your domain so it opens at yourdomain.com/llms.txt. The checker under the form tells you if anything is missing.
At the root, so it opens at yourdomain.com/llms.txt: that is the address Lighthouse requests, and the one the proposal names first. Version 2 also allows a file at a sub path, such as /docs/llms.txt, which covers the pages under that path. If your platform cannot serve a plain file at the root, ask your host or platform before spending time on the file.
Google says its Search does not use llms.txt, and that keeping one will neither harm nor help your visibility or rankings in Google Search. OpenAI's and Perplexity's crawler pages describe their robots.txt controls and do not say they read other sites' llms.txt files. A published study of 83 sites over 12 weeks (by a vendor of AI-visibility software) counted 7 requests for it from OpenAI's crawlers, 9 from Anthropic's ClaudeBot and none from PerplexityBot. So make one for the tools that do use it, not as a ranking tactic.
Three things: a line that starts with # and a space (an H1), at least one Markdown link, and at least 50 characters. A server error on /llms.txt, or a 429, fails the audit, and a plain 404 is marked not applicable because the file is optional. It is part of Lighthouse's experimental Agentic Browsing category, which needs Chrome 150 or later.
They do different jobs. The proposal says robots.txt tells automated tools what access is acceptable, such as for search indexing, while llms.txt is used on demand, when an agent needs information about a topic while helping a user. Neither replaces the other.
No. The file is built in your browser and nothing you enter, or paste into the checker, leaves the page.
Sources
- llmstxt.org The /llms.txt file, version 2 (Jeremy Howard, updated 10 August 2026)
- llmstxt.org Changes: what version 2 changed
- Google Search Central Optimizing your website for generative AI features on Google Search (last updated 10 July 2026)
- Chrome for Developers Lighthouse: the llms.txt audit
- Chrome for Developers Lighthouse: how the Agentic Browsing category is scored
- GoogleChrome/lighthouse on GitHub The llms-txt audit's source code (core/audits/agentic/llms-txt.js)
- OpenAI Overview of OpenAI crawlers
- Perplexity Perplexity crawlers
- EZY Research We put llms.txt on 83 websites. In 12 weeks, OpenAI's crawler read it 7 times (27 July 2026)
Every page above was opened on 20 November 2026. The proposal was last updated on 10 August 2026. Lighthouse’s audit and the log study are as published on that date and may change; the study is a vendor’s own and is labelled as one. The AI Chatbot facts come from its module page. ← All free tools
More on AI visitors and tracking
All tracking guides- How to Get Your Website on ChatGPT (and Track It) How to get your website into ChatGPT answers: let OpenAI's search crawler in, write pages it can quote, earn mentions, and track the visits. Free checker.
- How to Train a Chatbot on Your Own Website Content Train a chatbot on your website content: what training really means, what to add, how to write answers a bot can use, and a free check of what your text covers.
- Bot Traffic in Google Analytics: Spot It, Filter It GA4 drops known bots, but headless browsers, data-centre visits and spam still get in. The signs to look for, the filters that work, and a free checker.
- How to Track Website Traffic and See Where It Comes From Where is your website traffic coming from? How referrers, UTM tags and ad click IDs decide the source, where to look in GA4, and a free referrer decoder.
Stop losing the visitors
you already paid for.
More leads from the traffic you already have, and follow-up that sends itself. Free to start. No card, no monthly fee.