llms.txt: a developer's guide (with a real example)

llms.txt is a curated, Markdown map of your site written for language models rather than crawlers. It takes about ten minutes to ship — and this guide will also tell you the part most llms.txt guides leave out: what the evidence says it actually does in 2026, and where it belongs in your priorities.

What llms.txt is

llms.txt is a Markdown file at your site root — yourdomain.com/llms.txt — that gives language models a concise, curated map of your site: a one-line description of what you do, then your most important pages grouped into sections, each with a sentence of context. It’s a README for AI: instead of making an engine infer your site from navigation and marketing noise, you hand it the authoritative version.

What it does — and what it doesn’t

Most guides will tell you llms.txt is the highest-leverage move in AEO. The evidence says otherwise, and you deserve the real version before you spend time on it. Ahrefs analysed 137,000 sites and found roughly 97% of llms.txt files were never fetched by an AI bot. SE Ranking studied around 300,000 domains and found no significant correlation between having one and being cited in AI answers. Google has confirmed no Search system reads it, with John Mueller comparing it to the old keywords meta tag — a file the site owner controls, and therefore one engines have little reason to trust.

Where it genuinely earns its keep is AI developer tooling. Coding assistants like Claude Code, Cursor and Copilot fetch it to locate the right documentation without wasting context on your navigation and marketing copy. If you publish docs or an API, that alone is worth the ten minutes. It also costs nothing, can’t be penalised, and if adoption does arrive you’re already there.

So: ship it, but ship it last. The things that actually decide whether an answer engine cites you are letting AI crawlers through in robots.txt, structuring pages so the answer comes first and can be extracted, valid structured data — and, more than any on-site work, being mentioned on the third-party sources those engines already quote.

How to write one

The format is deliberately simple Markdown:

• An H1 — your site or brand name.
• A blockquote summary — one or two sentences on what the site is and who it's for.
• H2 sections of links — group your key pages (Core pages, Docs, Guides, Contact). Each link gets a short, factual description.
• Only what matters — this is a curated shortlist, not a sitemap. Link the pages you'd want an AI to read first.

A complete example

Here’s a full, valid llms.txt you can adapt. Save it as public/llms.txt (or serve it at your root) and update the names and URLs:

# Acme Analytics

> Acme Analytics is a privacy-first product analytics platform for
> SaaS teams. Track events, funnels, and retention without cookies
> or shipping user data to third parties.

## Core pages

- [Product overview](https://acme.example/product): what Acme does and who it's for
- [Pricing](https://acme.example/pricing): plans, limits, and the free tier
- [Docs](https://acme.example/docs): install, SDKs, and the events API
- [Security](https://acme.example/security): data handling, GDPR, and SOC 2

## Guides

- [Cookieless analytics, explained](https://acme.example/guides/cookieless)
- [Migrating from Google Analytics](https://acme.example/guides/ga-migration)

## Contact

- Sales: hello@acme.example
- Support: https://acme.example/support

That’s the whole thing. No schema to validate, no build step — just a clear, honest summary of your site. You can see a real one in production at traffix.dev/llms.txt. Want a head start? Our free llms.txt generator reads your site and drafts one for you.

Where it fits

llms.txt is the cheapest item in a larger AEO/GEO picture, not the most important one. Work the signals that decide citation first — AI-crawler access, answer-shaped content, valid structured data, and presence on the third-party sources engines already quote — then add llms.txt for the dev-tooling benefit. Our AEO/GEO guide covers the full checklist in priority order, and the free readiness check tells you which signals your site is missing right now.

Frequently asked

What is llms.txt?

llms.txt is a plain-text (Markdown) file at the root of your site — yoursite.com/llms.txt — that gives AI models a concise, curated map of what your site is and which pages matter. Think of it as a README for language models: instead of forcing an engine to infer your site from navigation and noise, you hand it an authoritative summary and the links worth reading.

Where do I put llms.txt?

At the web root, served at https://yourdomain.com/llms.txt with a text/plain or text/markdown content type. In a Next.js or static project, the simplest approach is a file in your public/ directory (public/llms.txt), which serves automatically at /llms.txt.

Does llms.txt actually do anything yet?

Be honest with yourself about this one: as of 2026, largely not for search. Ahrefs analysed 137,000 sites and found roughly 97% of llms.txt files were never fetched by an AI bot; SE Ranking looked at about 300,000 domains and found no significant correlation between having the file and being cited in AI answers; and Google has confirmed no Search system reads it. Where it does earn its keep is developer tooling — AI coding assistants like Claude Code, Cursor and Copilot fetch it to find the right docs without burning context. Add it for that, and because it's ten minutes and can't hurt you. Don't add it expecting citations.

Is llms.txt the same as robots.txt?

No. robots.txt tells crawlers what they may and may not access. llms.txt tells language models what your site is about and which content is most useful. They're complementary: robots.txt controls access, llms.txt provides context.

How is llms.txt different from a sitemap?

A sitemap is an exhaustive, machine-readable list of every URL for crawlers. llms.txt is a short, human-written, curated summary for language models — your most important pages with a sentence of context each, not every page on the site.