What Is llms.txt? The Format, Real Examples, a Template, and Who Reads It

llms.txt is a plain Markdown file that you put at the root of your website, at yoursite.com/llms.txt. It gives AI tools a short summary of your site and a list of links to your most useful pages. Think of it as a table of contents written for AI agents.

Jeremy Howard proposed it in September 2024 at llmstxt.org, and updated the proposal in August 2026. It is a proposal, not an official web standard. And as of October 2026, no major AI company has said it uses llms.txt to decide what to rank or cite.

What is llms.txt for?

llms.txt helps an AI agent find the right page on your site without reading all of it.

Web pages are built for people. They are full of menus, scripts and ads. An agent that wants one fact has to dig through all of that. Its context window, the amount of text it can hold at once, is too small for a whole website.

llms.txt solves that with one small file. It says what the site is, lists a few key facts, and links to clean versions of the important pages. The agent reads the file, follows only the links it needs, and skips the rest.

The proposal says these files are used most for software docs. A coding assistant that needs your API reference can find it in one hop.

The llms.txt format, line by line

The file uses Markdown, with a fixed order so programs can parse it too. Here are the parts, in order:

1. An H1 with the name of the site or project. This is the only required part.
2. A blockquote (a line starting with >) with a short summary.
3. Optional details: paragraphs or lists with key facts. No headings here.
4. Sections under H2 headings, each holding a list of links. Each item is a Markdown link, then an optional colon and a note.
5. A section named “Optional”, by convention, for links an agent can skip when it is short on space.

Here is the bare shape from the proposal:

# Title

> Optional description goes here

Optional details go here

## Section name

- [Link title](https://link_url): Optional link details

## Optional

- [Link title](https://link_url)

The August 2026 update added two ideas. A file can live in a folder, so /docs/llms.txt covers only the docs. And pages can offer a clean Markdown copy at the same address plus .md, so the links in llms.txt point to text an agent can read directly.

Real llms.txt examples

We opened these files on October 9, 2026. They show how differently teams use the same format.

Two patterns are worth copying. OpenAI keeps the top file tiny and splits the rest by product. Stripe puts its most important rules for agents right under the H1.

Our own llms.txt

Serplify publishes one at serplify.io/llms.txt. We generate it from the same data as our pages each time the site builds. So the prices in it always match the pricing page. We also publish serplify.io/llms-full.txt, which holds the full text of every guide and is built the same way. Here is the top of the file, trimmed:

# Serplify

> Serplify gives growth and SEO teams a Google SERP API, real website traffic, and real-click SEO traffic from one platform and one prepaid balance ...

Key facts, current as of this build:

- SERP API: $0.005 per Google results page (up to 100 results). Retries after a block are never billed.
- One prepaid balance covers all products and never expires. New accounts start with $1 of free balance.

## Products

- [Google SERP API](https://serplify.io/serp-api/): live Google results as JSON, endpoints, coverage and pricing.
- [Pricing](https://serplify.io/pricing/): per-result prices, top-up bonuses and what is billed when.

## Documentation

- [Quickstart](https://docs.serplify.io/getting-started/quickstart/): first SERP API request in curl, Node.js and Python.

The full file also lists every guide on this blog, grouped by topic, with its one-line summary.

Who reads llms.txt?

Mostly AI agents that are already on a task, not search engines. Here is what is known as of October 2026.

Google says you do not need it. Google’s guide to AI features in Search says you do not need new machine-readable files or “AI text files” to appear in AI Overviews or AI Mode. There is also no special schema markup to add.

Most files are never fetched. Ahrefs studied 137,210 sites that got traffic in May 2026. About 28 percent of them published an llms.txt file. Of those files, 97 percent got zero requests all month. Of the requests that did arrive, 96 percent came from bots. The top AI fetcher was GPTBot, OpenAI’s crawler, and the second was Claude Code, a coding agent.

Chrome checks for it. Lighthouse, Chrome’s site audit tool, has an llms.txt check in its agentic browsing audits. A missing file counts as “not applicable”. A file that returns a server error fails the check.

AI labs publish their own. OpenAI, Anthropic and Google’s Gemini team all keep an llms.txt for their developer docs. That tells you who the format helps most: developers whose coding tools read docs.

Put together, llms.txt is a convenience for agents. It is not a way to get into ChatGPT answers or Google’s AI Overviews. For that, see our guide to LLM SEO, and measure the result with AI visibility tracking.

llms.txt vs robots.txt vs sitemap.xml

They work together. robots.txt still decides whether AI crawlers can read your pages at all. A sitemap still helps Google find every page. If Google is missing pages, llms.txt will not fix it. Start with why your site may not show up on Google.

How to write an llms.txt file

1. Pick 10 to 30 pages. Choose the ones that answer what people most often ask about you: what you sell, prices, docs, policies.
2. Write the H1 and the summary. One or two plain sentences. Say what you do, for whom, and the fact people most often get wrong.
3. Add a short list of key facts. Prices with the month you checked them, limits, and what you do not do.
4. Group the links under H2 headings. Add a short note after each link so the agent knows when to open it.
5. Put extras under “Optional”. Blog feeds, long references, old docs.
6. Save it as plain text at /llms.txt. Use UTF-8. Serve it as text/plain or text/markdown.
7. Test it. The proposal suggests this: give an assistant only your llms.txt, then ask it questions about your site. If it gets them wrong, fix the file.
8. Keep it current. Build it from your site’s data, or use a tool. Yoast SEO, AIOSEO, Wix, Mintlify and GitBook can all generate one.

Here is a template you can copy:

# Your Company

> Your Company makes X for Y. One more sentence with the fact people most often get wrong.

Key facts:

- Price: $X per Y (checked October 2026).
- Works in: the countries or platforms you support.
- Does not: the thing people often assume you do.

## Products

- [Product name](https://example.com/product/): what it is, in one line.

## Docs

- [Getting started](https://example.com/docs/start/): first steps.
- [Pricing](https://example.com/pricing/): plans and what is billed.

## Optional

- [Blog](https://example.com/blog/): guides and news.

What about llms-full.txt?

llms-full.txt is a popular extra, not part of the core proposal. It holds the full text of all your docs in one Markdown file. Mintlify, for example, creates it on every docs site it hosts. Mintlify describes it as a file that combines a whole docs site into one.

It is useful when an agent needs everything at once, like a coding assistant loading a full API reference. It is usually too big for a normal website. The 2026 proposal leans the other way: small llms.txt files plus a Markdown copy of each page.

Should you add an llms.txt file?

Add one if you have docs, an API, or facts that AI tools often get wrong, like prices. It takes an hour, and a generated file stays correct on its own.

Do not expect traffic from it. The data so far says almost nothing fetches it, and Google does not use it for AI Overviews. Spend the bigger effort on pages that assistants cite: clear docs, clear pricing, and real answers to real questions. Then track whether Google’s AI Overviews cite you.