Is llms.txt an official standard?

Not yet. It was proposed by Jeremy Howard in September 2024 and has been voluntarily adopted by thousands of sites including Anthropic, Mintlify, and many documentation platforms. Major AI companies have not formally committed to reading it.

Does llms.txt actually improve AI citations?

The evidence is mixed. Sites that implement llms.txt often correlate with higher citation rates, but this likely reflects overall AEO maturity rather than llms.txt alone. It's a cheap addition that can't hurt.

How is it different from robots.txt?

robots.txt tells crawlers what they can access. llms.txt gives LLMs a structured, markdown-formatted map of your most important content — essentially a curated tour of your site optimized for AI consumption.

What is an llms.txt File and Should You Have One?

The short answer

llms.txt is a single markdown file placed at your site's root (/llms.txt) that tells large language models which pages are most important, what they contain, and how they're organized. It's not a blocking mechanism like robots.txt — it's a content guide.

What goes in it

The proposed spec has four sections:

Site title — H1 at the top
Description — a blockquote or paragraph summarizing what the site is about
Key content — markdown-style lists of important pages grouped by section
Optional — less critical content, delimited by an H2 "## Optional" section that LLMs can skip

Minimal example

# Example Co.

> Tools for answer engine optimization.

## Docs
- [Getting Started](/docs/start): First-time setup
- [API Reference](/docs/api): Full endpoint list

## Guides
- [AEO Basics](/guides/basics): What AEO is and why it matters

## Optional
- [Company history](/about): Background
- [Press kit](/press): Media assets

Who uses it today

As of early 2026, notable adopters include Anthropic, Vercel, Mintlify, Cloudflare Workers docs, Supabase, and thousands of open-source projects. Most adoption is concentrated in developer tools and documentation-heavy sites, but content marketers are starting to catch on.

You can browse the community-maintained directory at llmstxt.site for examples.

Does it actually work?

Honest answer: we don't know for sure yet. No major AI company has published a statement confirming they read llms.txt. Correlational data from AEO audits shows sites with llms.txt tend to rank higher on overall AEO scores — but those same sites also tend to have cleaner schema, better freshness signals, and stronger content structure. It's hard to isolate llms.txt's contribution.

That said, it's a 5-minute addition with zero downside. If AI engines start formally supporting it, early adopters benefit. If they don't, you've lost nothing.

How to create one

Three options:

Manual — write it yourself based on the spec. Fine for sites with 10-20 key pages.
Generator — use our llms.txt generator to build one from your sitemap in 30 seconds.
Framework plugin — many static site generators (Next.js, Astro, Nuxt) have community plugins that auto-build llms.txt from your content.

Host it at https://yoursite.com/llms.txt. That's it. No submission required — crawlers discover it automatically.