What Is llms.txt? The Honest 2026 Guide (+ How to Create One)

llms.txt is a markdown file you place at the root of your domain (/llms.txt) that gives AI systems a clean, curated map of your most important pages. Think of it as a robots.txt for large language models — but instead of telling crawlers what to avoid, it tells AI what's worth reading. It was proposed in September 2024 and adoption is growing, but there's a lot of hype around it. This guide covers what it actually is, how to create one, and the honest answer to whether it helps your SEO.

What is llms.txt?

llms.txt is a plain markdown file that lists your site's key content with short descriptions, so a language model assembling an answer can find the right pages quickly instead of crawling your whole site. It was proposed on September 3, 2024 by Jeremy Howard, co-founder of Answer.AI (and fast.ai).

The file lives at https://yourdomain.com/llms.txt. There's also a companion format, llms-full.txt, which inlines the full text of your docs into one file so an AI can read everything in a single fetch — handy for documentation sites.

It's deliberately simple: an H1 title, a one-line summary, and sections of markdown links. That's it. No special syntax, no server config — just a text file.

Does llms.txt help SEO or get you indexed by LLMs?

This is the question everyone asks, and most articles answer it dishonestly. Here's what the data actually says in 2026:

Google has said no, on the record. In July 2025, Google's Gary Illyes confirmed Google does not support llms.txt and isn't planning to. John Mueller compared it to the long-discredited keywords meta tag. Google has also stated llms.txt isn't needed for visibility in its AI Search features.
The citation bots barely read it. Across analyses of hundreds of millions of AI bot requests, the share of traffic from the crawlers that actually drive AI citations — GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot, Google-Extended — touching /llms.txt is statistically negligible.
Adoption is still modest. A study of 300,000 domains found roughly 10% have an llms.txt file — about one site in ten.

So the honest answer: in 2026, llms.txt does not improve your search rankings or your visibility in ChatGPT, Claude, Perplexity, or Gemini. Anyone selling it as a ranking or "GEO" factor is selling something the data doesn't support. If that's your only reason for adding one, you can skip it.

So what is llms.txt actually for?

It's not dead — it's just not an SEO tactic. The real, working use case is the agentic and developer-tool layer:

AI coding agents fetch it routinely. Cursor, Windsurf, Claude Code, GitHub Copilot, Cline, and Aider look for /llms.txt and /llms-full.txt when you point them at a documentation site, so they can pull accurate context instead of guessing.
Major companies publish one for exactly this. OpenAI, Anthropic, Stripe, Cloudflare, Vercel, and others ship llms.txt as a routing layer for AI agents working with their docs.

In other words, llms.txt is a Business-to-Agent (B2A) play, not an SEO one — the first standardized way to publish a machine-readable surface that AI agents and dev tools can route on. If you run a docs site, an API, a developer tool, or anything an AI assistant might help users build with, that's where it earns its keep.

How to create an llms.txt file

Creating one takes a few minutes. The format is:

Start with a single H1 title — your site or project name: # Your Project.
Add a one-line summary as a blockquote right after: > A short description of what your site is.
Group your key pages into ## Sections, each containing markdown links in the format - [Page name](https://url): short description.
Put a ## Optional section last for links an AI can safely skip if it's short on context.
Save it as llms.txt and upload it to the root of your domain so it's reachable at /llms.txt.

Keep it curated, not exhaustive — the whole point is to highlight what matters, not mirror your sitemap.

llms.txt format and example

A valid file is short and readable:

Markdown

# Example Project

> A one-sentence summary of the site for AI systems.

## Docs

- [Getting started](https://example.com/start): setup guide
- [API reference](https://example.com/api): full API reference

## Optional

- [Changelog](https://example.com/changelog): release history

That's the shape AI tools expect. Once you've written yours, it's worth confirming the structure is correct before you publish — a missing title or a malformed link can make the file unparseable. You can check it instantly with our free llms.txt validator, which scores the file and flags any errors.

How to view any website's llms.txt

Want to see a real one? Just add /llms.txt to any domain in your browser — for example https://stripe.com/llms.txt. If the site has published one, you'll see the raw markdown. It's a great way to study how mature teams structure theirs before writing your own; paste theirs into the llms.txt validator to see exactly how it's built.

Common llms.txt mistakes

The errors that break a file most often:

No H1 title, or more than one. Every file needs exactly one # title line.
Malformed links. A section with no links, or links not written as - [name](url): description.
Relative URLs that should be absolute. Use full https:// links.
A dumping ground instead of a curated map. Listing every page defeats the purpose.

Fix the errors, then run it through a validator to confirm a clean result before publishing.

FAQ

What is llms.txt supposed to be? A short, curated markdown map of your site's most useful pages, written for AI systems rather than human visitors. It's guidance, not a full content dump.

Is llms.txt the same as robots.txt? No. robots.txt tells crawlers what not to access; llms.txt tells AI which pages are worth reading. They solve opposite problems and can coexist.

What's the difference between llms.txt and llms-full.txt? llms.txt is a curated index of links. llms-full.txt inlines the actual content into one file so an AI can read everything in a single request.

Do I need an llms.txt file? If your goal is SEO rankings, no — it won't move them. If you run documentation, an API, or a developer product that AI coding agents help people use, yes — it makes your site agent-ready.