Quick answer: llms.txt is a Markdown file at https://your-domain.com/llms.txt that gives AI assistants a short, curated map of your most important pages. It starts with an # H1 name, a > one-line summary, and ## sections of links. Below: the official format, real examples, a copy-paste template, and an honest look at whether it helps.
What Is llms.txt?
llms.txt is a proposed standard for a plain-text Markdown file at your domain root that tells large language models (LLMs) and AI tools what your site is about and which pages matter most. Jeremy Howard of Answer.AI proposed it in September 2024; the specification lives at llmstxt.org.
The problem it solves: web pages are full of navigation, scripts and ads, and an AI tool answering a question has limited context. A short Markdown index of your best pages, each with a one-line description, is far easier for a model to use than crawling your whole site.
Some older guides and tools call it llm.txt (singular). The name in the proposal is llms.txt, so use that.
llms.txt Format (the Official Spec)
An llms.txt file is ordinary Markdown with a fixed order of sections:
- An H1 with the name of your site or project. This is the only required part.
- A blockquote summary (
> ...): one or two sentences with the key facts someone needs to understand the rest of the file. - Optional context: any number of paragraphs or lists (no headings) with extra detail, such as who you serve, regions, or how to interpret your pages.
- H2 sections of links (
## Products,## Docs, ...). Each item is a Markdown link, optionally followed by a colon and a short note:- [Title](https://url): note - An
## Optionalsection for secondary links that a tool can skip when it needs a shorter context.
That's the whole format. There are no Allow/Disallow rules, no key: value blocks and no priority fields. If a guide or tool shows you those, it isn't following the spec. Access control still belongs in robots.txt.
The proposal also suggests offering a clean Markdown version of important pages at the same URL with .md appended (for example /pricing.html.md), so AI tools can fetch the text without the HTML.
llms.txt Example (Copy-Paste Template)
Here is a complete example for a fictional online store. Notice how short it is: it's a curated brief, not a list of every URL.
# Acme Outdoor
> Acme Outdoor sells lightweight camping gear online and ships to the US and EU. This file lists the pages that best describe our products, policies and buying guides.
All prices are in USD. Every product includes a 2-year warranty.
## Products
- [Tents](https://www.example.com/collections/tents): 1-4 person backpacking tents with specs and weights
- [Sleeping bags](https://www.example.com/collections/sleeping-bags): Temperature ratings and fill types compared
## Buying guides
- [How to choose a backpacking tent](https://www.example.com/guides/choose-a-tent): Weight, capacity and season ratings explained
- [Tent size chart](https://www.example.com/guides/tent-sizes)
## Policies
- [Shipping](https://www.example.com/policies/shipping): Delivery times and costs by country
- [Returns](https://www.example.com/policies/returns): 60-day returns and how to start one
## Optional
- [About us](https://www.example.com/about)
- [Press](https://www.example.com/press)Blank llms.txt template
Copy this, replace the placeholders, and delete any section you don't need:
# Your Brand Name
> One or two sentences: what you do, who it's for, and what makes you different.
Optional context: pricing basics, regions you serve, or anything an AI should know before answering questions about you.
## Key pages
- [Main product or service](https://www.yourdomain.com/product): What it is and who it's for
- [Pricing](https://www.yourdomain.com/pricing): Plans and what's included
## Guides and docs
- [Getting started](https://www.yourdomain.com/getting-started): Setup steps
- [FAQ](https://www.yourdomain.com/faq): Answers to common pre-sale questions
## Policies
- [Shipping and returns](https://www.yourdomain.com/returns)
- [Contact](https://www.yourdomain.com/contact)
## Optional
- [Blog](https://www.yourdomain.com/blog/): Articles and announcementsReal llms.txt Examples
Developer-focused companies were the earliest adopters. These are live files you can open and study:
| Site | File | What it shows |
|---|---|---|
| Stripe | docs.stripe.com/llms.txt | Opens with instructions for AI coding agents before listing docs, so it shapes how assistants answer, not just what they read. |
| Cloudflare | developers.cloudflare.com/llms.txt | A top-level index where each product links to its own llms.txt. Useful pattern for large sites. |
| Perplexity | docs.perplexity.ai/llms.txt | Textbook format: H1, blockquote summary, then links pointing at .md versions of each page. |
| Vercel | vercel.com/llms.txt | Adds a "when to use Vercel" section, which helps assistants recommend the product in the right situations. |
| Anthropic | platform.claude.com/llms.txt | An overview of the Claude API documentation and developer resources. |
For e-commerce and SaaS brands, the most useful sections tend to be products or plans, pricing, policies (shipping, returns, warranty), comparison pages and FAQs: the pages that answer the questions buyers ask AI assistants.
Does llms.txt Actually Help?
Honest answer: it helps some AI tools, and it won't move your Google rankings.
- Google: Google representatives have said publicly that Google Search doesn't use llms.txt. AI Overviews draw on the normal search index, so classic SEO is what matters there.
- AI crawlers: OpenAI, Anthropic and Perplexity haven't documented their crawlers reading llms.txt for training or search. Don't expect a citation boost from the file alone.
- AI assistants and agents: this is where it's used today. Coding assistants, documentation tools, MCP servers and users pasting your URL into ChatGPT or Claude can fetch
/llms.txtto get a clean summary of your site instead of parsing HTML.
So treat llms.txt as a cheap, low-risk addition that takes 15 minutes, not as a ranking factor. What actually drives AI citations is the same work as good SEO: crawlable pages, clear structured data, content that directly answers buyer questions, and mentions on sites AI engines trust. For that broader playbook, see how to improve brand visibility in AI search engines.
llms.txt vs robots.txt vs sitemap.xml
| robots.txt | sitemap.xml | llms.txt | |
|---|---|---|---|
| Purpose | Which bots may crawl which paths | Every URL you want indexed | A short, curated brief of your best pages |
| Format | Plain-text directives | XML | Markdown |
| Main readers | All crawlers, including GPTBot and ClaudeBot | Search engines | AI assistants, agents and tools |
| Size | A few lines | Up to 50,000 URLs | Usually 10-100 links |
Use all three. If your robots.txt blocks AI crawlers, publishing an llms.txt won't undo that. Our guide to GPTBot, ClaudeBot and PerplexityBot covers those rules.
What Is llms-full.txt?
llms-full.txt is a companion convention: one Markdown file containing the full text of your key pages, so a tool can load everything in one request instead of following links. It's popular on documentation sites (many docs platforms generate both files automatically). It isn't part of the core proposal, and for most marketing sites and stores a well-written llms.txt is enough.
How to Create and Publish llms.txt
- Draft it. Use the template above or the free llms.txt generator, which builds a spec-compliant draft from your sitemap.
- Curate it. Cut it down to the pages you'd want an assistant to quote. Write a useful note after each link; "Pricing" alone tells a model little, "Pricing: plans from $9/month, annual discount" tells it a lot.
- Upload it to your root so it loads at
https://your-domain.com/llms.txt.- Static sites, Next.js, Vercel, Netlify: put the file in your public or root folder.
- WordPress: upload via FTP, or publish it from admin with the free SellOnLLM WordPress plugin.
- Shopify: Shopify doesn't let you upload files to the domain root directly, so use an app such as our Shopify app.
- Verify it. Open the URL in a browser: it should return the raw text (not a 404 or an HTML page). The free AEO/GEO checker Chrome extension confirms it's found and checks your robots.txt rules for AI crawlers at the same time.
Common llms.txt Mistakes
- Dumping every URL. That's your sitemap's job. A 2,000-link llms.txt defeats the purpose.
- Inventing directives. Lines like
priority: highorDisallow:aren't part of the format and are ignored at best. - Links without notes. The short description after each link is what helps a model choose the right page.
- Letting it go stale. Redirected or deleted URLs undermine trust. Review it when you launch or retire key pages, or at least quarterly.
- Blocking AI crawlers in robots.txt while expecting AI visibility.
Does Every Site Need llms.txt?
No, but it costs almost nothing. It's most useful for sites with documentation or APIs (developers' AI tools read it constantly), for stores and SaaS products with many pages where a curated brief helps assistants find pricing, policies and comparisons, and for any brand that wants a clean, quotable summary of itself available to AI tools.
Summary
- What it is: a Markdown file at
/llms.txtwith an H1 name, a blockquote summary and H2 sections of annotated links. - Who reads it: AI assistants, agents and developer tools. Google Search doesn't use it.
- How to make one: copy the template above or use the free generator, curate it, upload it to your root, and verify it loads.