
The Website Owner’s Complete Guide to llms.txt: What It Is, How It Works & How to Generate Yours for Free
There is a quiet revolution happening on the web, and most website owners have no idea it is taking place. A new file - one so small you could write it in a lunch break - is being added to thousands of websites every single day by some of the most technically sophisticated companies on the planet. Anthropic, Cloudflare, Stripe, Perplexity, and NVIDIA have all quietly added it to their sites. Yoast, AIOSEO, and RankMath have built generators for it. And the broader SEO community is calling it the single most important website tweak of the decade.
The file is called llms.txt, and by the time you finish reading this guide, you will understand exactly what it is, why it matters enormously for your website’s visibility in the age of artificial intelligence, and precisely how to create yours for free in under 30 minutes.
Quick Summary
llms.txt is a lightweight plain-text file you place at the root of your website. Its sole purpose is to help AI tools - ChatGPT, Claude, Perplexity, Gemini, and others - quickly understand what your website is about and which pages matter most. Think of it as a crisp, well-organised briefing note written specifically for artificial intelligence.
1. What Exactly Is llms.txt?
llms.txt is a plain-text file written in Markdown format, placed at the root of your website so that it is accessible at the address yourwebsite.com/llms.txt. Its entire purpose is to give AI language models a clean, structured, and easy-to-read summary of your website: what you do, who you serve, and which pages contain your most important content.
The name itself is straightforward. LLMs stands for Large Language Models - the AI systems that power tools such as ChatGPT, Claude, Gemini, and Perplexity. The .txt extension follows the same naming convention as the classic robots.txt file that webmasters have relied upon for decades. Just as robots.txt was created to help search engine crawlers navigate your site sensibly, llms.txt was designed to help AI models understand your site intelligently.
In the simplest possible terms: llms.txt is your website’s introduction letter to the artificial intelligence world.
| File | Talks To | Purpose |
|---|---|---|
| robots.txt | Google, Bing, search crawlers | Controls which pages crawlers can index or ignore |
| sitemap.xml | Search engines | Lists every URL on your website |
| llms.txt | ChatGPT, Claude, Perplexity, Gemini | Tells AI what your site is about and highlights key pages |
| llms-full.txt | AI language models | Contains your entire website content in one readable file |
2. Who Created llms.txt? The Origin Story
llms.txt was created by Jeremy Howard, co-founder of Answer.AI and fast.ai, and a lecturer at the Universities of Queensland and Stanford. He formally published his proposal on 3rd September 2024, simultaneously releasing the official specification website at llmstxt.org and detailing his thinking on the Answer.AI blog at answer.ai.
Howard’s insight came from a deceptively simple observation. When AI tools attempt to read a typical webpage, they are immediately overwhelmed by noise: navigation menus, cookie-consent banners, JavaScript code, advertising scripts, sidebar widgets, and footer links all compete for space inside an AI’s limited “context window.” The genuinely useful content - the thing the AI actually came for - gets buried beneath layers of digital clutter.
He noticed that robots.txt had already solved a comparable problem for search engines back in the 1990s by giving websites a simple, standardised way to communicate with automated crawlers. His question was straightforward: why not create the same kind of file for AI language models?
The result was llms.txt - a clean, minimal, human-readable and AI-readable file that cuts through all the noise and hands AI tools precisely what they need to understand your website, nothing more, nothing less.
Official Resources
- llmstxt.org - The official specification and proposal
- answer.ai - Jeremy Howard’s original announcement post
- fast.ai - Jeremy Howard’s AI research organisation
3. Is llms.txt an Official Web Standard? The Honest Answer
This is one of the most frequently searched questions about llms.txt, and it deserves a straight, unvarnished answer: llms.txt is a community-proposed standard, not an official W3C or IETF web standard. As of early 2026, no major AI company has publicly confirmed that their systems actively read or prioritise llms.txt files during live inference.
However, this is precisely where things become fascinating. Despite the absence of any formal industry endorsement, the organic adoption of llms.txt has been extraordinary. Over 844,000 websites had already implemented it by late 2025, according to BuiltWith tracking data. Major technology companies including Anthropic, Cloudflare, Stripe, Perplexity, Hugging Face, Cursor, Supabase, Zapier, ElevenLabs, and NVIDIA all have llms.txt files live on their websites right now.
The most significant adoption milestone came in November 2024, when Mintlify - one of the most widely used developer documentation platforms - enabled automatic llms.txt generation for every documentation site it hosts. Practically overnight, thousands of technical documentation sites, including those belonging to Anthropic and Cursor, began supporting the standard without their owners lifting a finger.
| Status Item | Detail |
|---|---|
| Proposed by | Jeremy Howard, Answer.AI - 3rd September 2024 |
| Specification hosted at | llmstxt.org (community-maintained, not W3C) |
| Official W3C / IETF standard? | No - community proposal only at present |
| Major AI platform confirmation? | Not officially confirmed, yet widely adopted |
| Number of sites using it | 844,000+ as of October 2025 |
| CMS and plugin support | Yoast, AIOSEO, RankMath, Webflow, Shopify apps |
4. How Does llms.txt Actually Work?
Understanding how llms.txt works requires a brief look at how modern AI tools retrieve information from the web. When you ask ChatGPT, Perplexity, or Claude a question about a particular company or topic, the AI often goes out to the web in real time to find current, accurate information. This process is known as inference-time retrieval.
The problem is this: when an AI visits a typical webpage, it must parse through vast quantities of HTML code. Every navigation link, sidebar widget, pop-up banner, JavaScript function, CSS stylesheet, and footer element gets loaded into the AI’s context window, consuming valuable space that could be used for actual content. On a mid-sized website, this HTML noise alone can exceed 500,000 tokens - far more than most models can process efficiently.
llms.txt solves this by providing AI tools with a single, clean starting point. Here is the step-by-step process:
- An AI tool or LLM-powered assistant visits your website to answer a user’s query.
- Before crawling individual pages, the AI checks your root directory for a file called llms.txt.
- If found, the AI reads this lightweight Markdown file first, gaining an immediate, organised overview of your entire website.
- The AI then selectively fetches only the specific pages it requires, guided by the structured links in your llms.txt.
- This dramatically reduces token usage and delivers clean, noise-free content to the model.
- The result is faster, more accurate, and more relevant responses about your website and its content.
The Two-File System: llms.txt and llms-full.txt
- llms.txt: The navigation file. Short, curated, and structured. It acts like a table of contents, directing AI to your most important pages. This is the file most website owners need.
- llms-full.txt: The comprehensive file. This contains your entire website content flattened into a single document. Cloudflare’s version contains 3.7 million tokens. Vercel’s is essentially 400,000 words of content. The logic is simple: give the AI everything upfront so it never needs to follow links at all.
5. The llms.txt Format: Structure, Syntax, and a Ready-to-Use Template
One of the most appealing things about llms.txt is its elegant simplicity. It uses standard Markdown formatting, which is already the native language of most large language models. There is no XML schema to learn, no JSON configuration to wrestle with, and no complex setup required — just clean, readable text.
Here is the official structure as defined by the specification at llmstxt.org:
Required Elements
- H1 Header (#): Your project or website name. This must be the very first element in the file.
- Blockquote (>): A one-to-three-sentence summary of what your website does. This often becomes the AI’s core mental model of your brand.
Strongly Recommended Elements
- Additional context paragraphs: Further detail about who you serve, what problems you solve, and any important instructions for AI tools reading the file.
- H2 Sections (##): Grouped categories of links — for example, Documentation, Blog, API Reference, Tools, Getting Started.
- Link lists: Each link follows the format: [Page Title](URL): Brief description of what is on this page.
Optional Elements
- An Optional section for secondary or lower-priority resources.
- Specific instructions for AI tools about how to cite or use your content.
Full llms.txt Template — Copy and Customise
One to three sentence summary of what your website is about, who you serve, and what makes you unique or valuable.
Additional context about your organisation goes here. You can explain your core products or services, your target audience, your unique value proposition, or any important notes for AI tools reading this file. Keep it concise, factual, and authoritative.
Getting Started
- Home Page – https://yoursite.com/ – Main landing page and overview
- About Us – https://yoursite.com/about – Company background and mission
- Getting Started Guide – https://yoursite.com/start – Step-by-step for new visitors
Documentation
- Full Documentation – https://yoursite.com/docs – Complete product documentation
- API Reference – https://yoursite.com/api – All API endpoints and usage examples
- FAQ – https://yoursite.com/faq – Frequently asked questions
Blog
- Blog – https://yoursite.com/blog – Latest articles and tutorials
- Case Studies – https://yoursite.com/cases – Real-world implementation examples
Tools
- Free Tool Name – https://yoursite.com/tool – Brief description of what this tool does
Optional
- Privacy Policy – https://yoursite.com/privacy – How we handle your data
- Contact – https://yoursite.com/contact – How to get in touch with us
Real-World Example: Anthropic’s llms.txt
Anthropic, the company behind the Claude AI assistant, has one of the most widely cited llms.txt implementations on the web. Their file at anthropic.com/llms.txt contains 8,364 tokens and links methodically to all their key API documentation, developer guides, and product resources. Their companion llms-full.txt contains 481,349 tokens of complete documentation content. This is widely regarded as the gold standard for how enterprise organisations should approach the specification.
6. Why Every Major Tech Company Is Talking About llms.txt
The reason llms.txt has gone viral across the tech and SEO world is not simply about the technology itself. It is about what the technology represents: a fundamental, irreversible shift in how the internet works and how people find information.
For the past 25 years, websites have been built for human readers and optimised for Google. The entire SEO industry exists to help pages rank in search results so that human users click on them. But a new reality is emerging rapidly. AI tools such as ChatGPT, Perplexity, and Claude are increasingly becoming the first port of call for people seeking answers. And these AI tools do not work like Google.
The Numbers Driving the Conversation
- AI models including ChatGPT, Claude, Grok, and Gemini are estimated to be driving between 10 and 30 per cent of web search traffic, according to Mintlify and Yotpo forecast data from 2025.
- Over 844,000 websites had implemented llms.txt by October 2025, up from effectively zero in September 2024.
- One case study from Springs Apps reported a 20 per cent increase in search engine visibility and a 15 per cent improvement in accurate AI-driven query responses after implementing llms.txt.
- Reducing AI hallucinations about your brand by anywhere between 30 and 70 per cent is a claimed benefit when models have clean, structured llms.txt data to reference rather than attempting to infer from cluttered HTML.
Which Major Companies Have Already Implemented llms.txt?
| Company | Why It Is Noteworthy |
|---|---|
| Anthropic (Claude) | Has both llms.txt and llms-full.txt. Their 8,364-token navigation file is considered a benchmark implementation. |
| Cloudflare | Organises their file neatly by product area. Their llms-full.txt runs to 3.7 million tokens. |
| Stripe | Structured by product categories with a dedicated Optional section for specialist tools. |
| Perplexity | Particularly telling: the AI search engine itself uses llms.txt to guide other AI tools. |
| Hugging Face | One of the earliest major adopters in the AI research community. |
| Cursor | Developer tool that benefits directly from AI coding assistants reading structured docs. |
| Supabase | Open-source database platform helping AI coding tools understand their documentation precisely. |
| Zapier | Automation platform using llms.txt to help AI tools understand thousands of integrations. |
| ElevenLabs | Voice AI company ensuring AI tools accurately reference their capabilities and pricing. |
| NVIDIA | Split implementation covering both technical documentation and the main corporate site. |
Jeremy Howard himself articulated the broader stakes clearly in early 2025 when he observed that the vast majority of content on the web is still written for humans, whilst the overwhelming majority of future ‘attention’ — the queries, the retrieval requests, the information lookups — will come from AI models rather than human eyeballs. Whether his precise timeline proves accurate, the direction of travel is unmistakable.
7. Is llms.txt Actually Useful? An Honest Assessment
This is the question every website owner deserves a candid answer to. The truth is: the supporting evidence is still accumulating, but the cost of implementing llms.txt is so negligible that the risk-to-reward calculation overwhelmingly favours doing it.
The Case in Favour of llms.txt
- Zero downside risk: It is a single plain-text file. It costs nothing to create, takes 10 to 30 minutes to set up, and carries absolutely no technical risk whatsoever.
- Early mover advantage: Fewer than one million websites have implemented it out of the billions that exist online. Early adoption gives you a clear positioning advantage as AI-driven discovery continues to grow.
- Brand accuracy: When AI tools have a structured, authoritative source coming directly from you, they are far less likely to hallucinate incorrect information about your products, services, or company history.
- Developer and technical audience: If your audience includes developers using AI coding tools such as Cursor or GitHub Copilot, llms.txt helps those tools reference your documentation with precision.
- Ecosystem momentum: With Yoast, AIOSEO, RankMath, and Webflow all adding native llms.txt support, the technology is being treated as a genuine standard by the industry’s leading tools.
The Case for Caution
- No confirmed AI platform support: As of early 2026, not a single major AI company has publicly stated that their systems actively read llms.txt files during inference. The file may well be used, but it has not been confirmed.
- Not a Google ranking factor: llms.txt has no known impact on traditional Google search rankings. It should complement your existing SEO strategy, not replace it.
- Still a proposal: The specification lives on a community website rather than with W3C or IETF. It may evolve, change, or be superseded by something else entirely.
Verdict
Implement llms.txt. It takes less time than writing a single blog post, costs nothing, and positions your website sensibly for the AI-first web that is clearly on its way. Simply do not treat it as a replacement for your established SEO efforts — treat it as a smart, low-effort addition to them.
8. How to Add llms.txt to Your Website: Step-by-Step
Whether you run a WordPress blog, a Shopify store, a bespoke-coded site, or a SaaS product, adding llms.txt is straightforward. Here are your three options:
Option A: Create It Manually
- Open any plain-text editor (Notepad on Windows, TextEdit in plain-text mode on Mac, or VS Code).
- Write your llms.txt content following the format template in Section 5 above.
- Save the file as exactly llms.txt with no other extension.
- Upload it to the root directory of your website via FTP, cPanel File Manager, or your hosting control panel.
- Verify it is live by visiting yoursite.com/llms.txt in a browser.
Option B: Use a WordPress Plugin
- Yoast SEO: Now generates llms.txt automatically from your existing sitemap. Update Yoast to the latest version and look for the llms.txt toggle within settings.
- AIOSEO (All in One SEO): Over 3 million active users. Has built-in llms.txt generation with granular control over which pages to include or exclude.
- RankMath: Popular WordPress SEO plugin with llms.txt support added in a recent update.
Option C: Use a Free Online Generator
The fastest option for most website owners. You simply paste your URL, and the tool crawls your site and generates your llms.txt file in under 60 seconds. All the tools in the next section are entirely free.
9. The Best Free llms.txt Generator Tools Available Right Now
You do not need a developer, a plugin, or a paid subscription to create your llms.txt file. Every tool listed below is free, and most require no sign-up or credit card whatsoever.
| Tool | URL | Best For | Cost |
|---|---|---|---|
| llmstxtgenerator.org | llmstxtgenerator.org | General websites, fast generation | Free |
| SiteSpeak AI Generator | sitespeak.ai/tools/llms-txt-generator | No sign-up needed, 30-second results | Free |
| Writesonic Generator | writesonic.com/free-tools/llms-txt-generator | Content marketers and SEO professionals | Free |
| LLMrefs Generator | llmrefs.com/tools/llms-txt-generator | Deep site crawl with AI-written descriptions | Free |
| WordLift Generator | wordlift.io/generate-llms-txt | Knowledge graph and semantic SEO users | Free |
| llms-txt.io | llms-txt.io | Generates both llms.txt and llms-full.txt | Free |
| Keploy Generator | keploy.io/llmstxt-generator | Developers, no login required | Free |
| Rankability Generator | rankability.com | WordPress and non-WordPress, sitemap upload | Free |
| Seomator Generator | seomator.com/free-llms-txt-generator | SEO professionals, broader tool suite | Free |
| LiveChatAI Generator | livechatai.com/llms-txt-generator | Businesses with AI chatbot integrations | Free |
How to Use Any Generator: Universal Steps
- Navigate to any generator URL from the list above.
- Paste your website URL into the input field.
- Click Generate and allow 15 to 60 seconds for the tool to crawl your site.
- Review the output and refine any descriptions that feel inaccurate or generic.
- Copy the text or download the file directly.
- Save the file as llms.txt and upload it to your website’s root directory.
- Confirm it is working by visiting yourwebsite.com/llms.txt in your browser.
10. Common Mistakes to Avoid
- Wrong file location: Your llms.txt must sit at the root of your domain at yoursite.com/llms.txt, not buried in a subfolder or subdirectory.
- Using HTML formatting: The file must be plain-text Markdown only. No HTML tags, no styling attributes, no JavaScript of any kind.
- Listing every page on your site: Be selective and deliberate. Only link to your most important, highest-quality pages. Quality and curation matter far more than volume.
- Setting it up and forgetting it: As your site grows and evolves, keep your llms.txt updated. A quarterly review is a sensible habit to build.
- Treating it as a replacement for robots.txt: These files serve entirely different purposes and you need both. robots.txt controls crawler access permissions; llms.txt provides content context and guidance.
- Expecting overnight SEO results: llms.txt is a long-term positioning move for the AI-driven web, not a quick-win ranking hack.
11. The Future of llms.txt and Generative Engine Optimisation
The broader movement that llms.txt represents — often referred to as GEO, or Generative Engine Optimisation — is still in its earliest stages. As AI tools such as ChatGPT Search, Perplexity, Google AI Overviews, and Claude continue to handle an ever-increasing share of the world’s information queries, the question of how websites communicate with AI systems will become as strategically important as traditional search engine optimisation is today.
Whether llms.txt becomes the definitive long-term standard or eventually evolves into something else, the underlying principle it embodies is here to stay: websites that make themselves easy for AI to understand, parse, and reference accurately will hold a significant competitive advantage in the AI-first web that is fast approaching.
The organisations implementing llms.txt today — Anthropic, Cloudflare, Stripe, Perplexity, and hundreds of thousands of others — are not acting on proof that it works. They are acting on the near-certainty that the cost of doing so is negligible and the potential upside is considerable.
Take Action: Add llms.txt to Your Website Today
Add llms.txt to your website today. Use one of the free generator tools listed in Section 9. It will take you less than 30 minutes and it positions your website intelligently for the inevitable AI-driven future of search and information discovery. There is genuinely no reason not to.


