llms.txt example and format: how to write one that passes
An llms.txt file is a short Markdown file at the root of your site that tells AI assistants what your site is and where the useful pages are. Jeremy Howard proposed it on September 3, 2024, and published version 2 of the proposal on August 10, 2026 at >llmstxt.org.
This guide shows the format, a copy-ready example, and the mistakes our llms.txt checker catches most.
What llms.txt is for
Web pages are built for people. Navigation, ads, and JavaScript make them hard for an AI agent to read. The llms.txt proposal gives agents one place to find concise information and links to clean versions of your key pages.
The v2 proposal says thousands of sites publish an llms.txt file, that documentation platforms generate one, and that Chrome's Lighthouse audits sites for one as part of its agentic browsing checks. It names OpenAI, Anthropic, and Gemini as publishers of llms.txt files for their own developer docs.
llms.txt does not replace robots.txt or your sitemap:
- robots.txt says which crawlers may access which paths. See our AI crawler robots.txt guide.
- sitemap.xml lists every indexable page for search engines.
- llms.txt is a curated overview that an agent reads on demand when it helps a user.
Where the file goes
Put it at /llms.txt on your site, for example https://example.com/llms.txt. The v2 spec allows a file at a subpath, besides the root one. A file such as /docs/llms.txt covers the pages under /docs/. When more than one file applies, agents should use the most specific one.
The format, in order
The spec lists these sections, in this order:
- An optional byte-order mark.
- An H1 with the name of the site or project. This is the one required section.
- A blockquote with a short summary that holds the key facts needed to understand the rest of the file.
- Zero or more Markdown sections of any type except headings (paragraphs, lists) with more detail.
- Zero or more sections under H2 headings, each a "file list" of links. Each list item is a Markdown link
[name](url), followed by an optional colon and notes.
By convention, an H2 named Optional holds secondary links an agent can skip when it needs a shorter context.
A copy-ready example
This file is for a made-up plumbing company. Swap in your own details.
# Riverside Plumbing
> Licensed plumbing company serving Springfield, IL and nearby towns since 2009. We fix leaks, clear drains, and install water heaters. 24/7 emergency service. Call (555) 010-0199.
Service area: Springfield, Chatham, Rochester, and Sherman. We do not serve commercial high-rises.
## Services
- [Water heater installation](https://riversideplumbing.example/water-heaters): Tank and tankless, same-week installs
- [Drain cleaning](https://riversideplumbing.example/drains): Snaking and camera inspection
- [Emergency plumbing](https://riversideplumbing.example/emergency): What to do before we arrive
## Company
- [Pricing](https://riversideplumbing.example/pricing): Trip fee and common job prices
- [Reviews](https://riversideplumbing.example/reviews)
## Optional
- [Blog](https://riversideplumbing.example/blog)
That file has the required H1, a summary with the facts an assistant needs to answer "who is this and where do they work," and short notes on each link.
Markdown versions of your pages
The proposal goes one step further. It suggests that pages agents might need offer a clean Markdown copy at the same URL with .md added, such as page.html.md, or with the extension replaced, such as page.md. URLs without a file name would use index.html.md or index.md.
To help agents find these files, v2 recommends standard link relations: rel="alternate" type="text/markdown" for the Markdown copy, and rel="describedby" for the llms.txt file that covers the page. You can send these as HTML <link> tags or as an HTTP Link: header.
This part is optional. A plain llms.txt with good links helps on its own.
Five mistakes to avoid
- No H1. The H1 is the one required element. A file that opens with a paragraph or an H2 fails the format.
- Headings inside the detail block. Between the summary and the first H2, the spec allows any Markdown except headings. An H3 in that block breaks the structure.
- Bare URLs in file lists. Each list item should be a Markdown link with a name, not a raw URL.
- Wrong place or wrong type. The file must load at
/llms.txtover HTTPS with no login. A redirect to your homepage, or an HTML error page served with a 200 status, does not count. - A dump of every page. The spec says the file should stay small to fit in an AI's context. Put the 10 to 30 pages that matter, not your whole sitemap.
Paste your file into our llms.txt validator to check the format line by line.
How common is llms.txt?
On October 17, 2026 we checked 49 popular SaaS websites. 35 of them served an llms.txt file. The 14 that did not included Salesforce, Trello, Airtable, GitLab, Figma, Canva, Zoom, and DocuSign. The full table is in the State of AI readiness report and as JSON.
Test your file
- Publish the file at
/llms.txt. - Run the AI readiness check on your domain. It tests llms.txt along with AI crawler access, JSON-LD, sitemap, meta description, and text readable without JavaScript.
- Follow the spec's own advice: ask an AI assistant questions about your business, giving it your llms.txt as the starting point, and judge the answers.
- Add the AI readiness badge to your site after you pass.
FAQ
Is llms.txt an official standard? It is a proposal, published at llmstxt.org, that is open for community input. It is not an IETF or W3C standard.
Does llms.txt help Google rankings? The proposal is about helping AI agents read your site. It makes no claim about search rankings.
What is llms-full.txt? Some sites publish a second file with the full text of their docs in one place. The llmstxt.org page centers on llms.txt and Markdown page copies.
Do I need llms.txt if I block AI crawlers? llms.txt helps agents that are allowed to read your site. If robots.txt blocks them, fix that first with the AI crawler robots.txt checker.
Checked 2026-10-17. Prices and policies change. Confirm on the source before you buy.