What Is llms.txt? Format, Example, and Whether You Need One

llms.txt is a plain Markdown file at the root of your site (/llms.txt) that tells large language models what your site is and where its most useful pages are. Think of it as a short, curated table of contents written for AI assistants rather than for search engine crawlers. It was proposed in September 2024 by Jeremy Howard of Answer.AI and has spread quickly through developer documentation sites.

This guide explains what llms.txt is, the exact llms.txt format, a complete llms.txt example you can adapt, and the honest answer to the question everyone asks: does anything actually read it?

What problem llms.txt tries to solve

When an AI assistant fetches your site to answer a question, it gets HTML full of navigation, cookie banners, scripts and layout. The useful content is buried, and the model's context window is limited. llms.txt gives the model a shortcut: a clean summary of the site and a short list of the pages worth reading, in a format models handle well. The idea is the same one behind good documentation: tell the reader where to start.

llms.txt vs robots.txt vs sitemap.xml

The three files are often confused, but they do different jobs:

llms.txt does not grant or block access. If robots.txt blocks an AI crawler, an llms.txt file will not help it. Get robots.txt right for AI crawlers first; our robots.txt and sitemap guide covers both files in depth.

The llms.txt format

The llms.txt format is deliberately simple, and the order of sections matters:

1. An H1 with the site or project name. This is the only required element.
2. A blockquote with a one or two sentence summary of what the site is.
3. Optional paragraphs with anything a model should know: who the product is for, key terms, what not to assume.
4. H2 sections containing Markdown lists of links, each in the form [Title](URL): short description.
5. An optional section named "Optional" for secondary links a model can skip when its context is tight.

An llms.txt example

Here is a complete llms.txt example for a fictional SaaS product. Copy it, replace the content, and keep it short: the point of llms.txt is curation, not completeness.

# Acme Analytics

> Acme Analytics is privacy-friendly web analytics for small businesses.
> No cookies, GDPR-compliant by default, one script under 2 KB.

Acme is a hosted service. Pricing is per site, not per pageview.
The open-source tracker lives on GitHub; the dashboard is closed source.

## Product

- [Features](https://acme.example/features): what Acme tracks and what it never collects
- [Pricing](https://acme.example/pricing): plans, limits and the free tier
- [Security](https://acme.example/security): data residency, encryption and compliance

## Docs

- [Quick start](https://acme.example/docs/start): add the script and verify tracking
- [Events API](https://acme.example/docs/events): send custom events from your backend
- [Migrating from Google Analytics](https://acme.example/docs/migrate-ga): importing history

## Optional

- [Changelog](https://acme.example/changelog): release notes
- [Blog](https://acme.example/blog): product updates and guides

Serve it as text/plain or text/markdown, not as an HTML page. A common mistake is a framework catch-all route that returns your 404 page or homepage HTML at /llms.txt with a 200 status. That looks like an llms.txt file to a quick check but is useless to a model.

What about llms-full.txt?

Some sites also publish llms-full.txt, which contains the full text of the key pages in a single Markdown file, so an agent can load everything in one request. It is popular with developer documentation (it is handy to paste into a coding assistant). It can get large; if you publish one, generate it from your docs at build time so it never goes stale.

Does anything actually read llms.txt?

Here is the honest picture as of late 2026:

• Google Search does not use llms.txt. Google's search team has said publicly that it is not used for ranking or for AI Overviews. Do not expect a traffic bump in Google from adding one.
• The major AI assistants have not committed to it. No large AI company has documented that its crawler reads llms.txt as a ranking or citation signal. Server logs on sites that publish one show only occasional fetches.
• Developer tools do use it. Coding assistants, documentation tools and MCP-based doc servers read llms.txt and llms-full.txt to pull accurate docs into an agent's context. If you run a developer product, this is the strongest reason to publish one.

So should you add llms.txt? It costs an hour and does no harm, and for documentation-heavy and developer-facing sites it already helps. For a local business or an ecommerce store, the things that move AI search visibility are still crawl access, clear content and structured data. Treat llms.txt as a low-cost extra, not a strategy.

How to create an llms.txt file

1. List the 5 to 20 pages that best explain your product or site. Start with what a new customer or a new developer would read first.
2. Write the H1 and a two-sentence summary. Be plain and specific; this is the text a model is most likely to quote.
3. Group the links under a few H2 headings, each with a one-line description.
4. Save it as llms.txt in your public root (for example public/llms.txt in Next.js, Astro or Vite projects).
5. Check that https://yoursite.com/llms.txt returns status 200, a non-HTML content type, and starts with a Markdown heading.
6. Review it whenever your main pages change. A stale llms.txt that points at removed pages is worse than none.

How Scanverra checks llms.txt

The Scanverra SEO Optimizer checks whether your site serves an llms.txt file at the root, and it is strict about it: the file must return a successful status, must not be an HTML page, and must start with a Markdown heading, so a framework fallback page does not count. Because no search engine uses llms.txt, Scanverra reports it as information only and never lowers your score for a missing one. The same scan checks which AI crawlers your robots.txt allows (search bots such as OAI-SearchBot, Claude-SearchBot and PerplexityBot versus training bots such as GPTBot and Google-Extended), along with structured data and the rest of your on-page SEO. Those crawler settings matter far more for AI visibility than llms.txt does today.

Key takeaways

• llms.txt is a curated Markdown map of your site for AI models, served at /llms.txt.
• The llms.txt format is an H1, a blockquote summary and sections of links.
• It does not replace robots.txt or sitemap.xml, and Google does not use it.
• It is worth publishing for developer and documentation sites, and harmless elsewhere.