# What is llms.txt, and do AI engines read it?

> llms.txt is a proposed standard: a Markdown file at a site's root that lists its most useful pages, with a line on each, for language models to read. Proposed in 2024, it is used mostly by coding agents reading software docs. No AI answer engine has said it reads the file, so it is cheap to add but not a fix. Letting AI crawlers in and serving content without JavaScript matter more.

| | |
| --- | --- |
| Where | /llms.txt at the site's root |
| Format | Markdown: a title, a summary, then lists of links |
| Proposed | 2024, llmstxt.org (now version 2) |
| Manifold tool | aeo_get_site_readiness checks it, free |

## What goes in the file

The [proposal](https://llmstxt.org/) sets a simple format:

1. An H1 with the site's name, the only required part.
2. A short summary in a blockquote.
3. Optional notes.
4. Sections under H2 headings, each a list of links with a line on what each page holds.

It also proposes a Markdown version of each page at the same URL plus `.md`, so an agent gets clean text instead of HTML.

## Do AI engines read it?

The proposal says the files are used most by coding agents looking up software documentation. No AI answer engine has documented reading `llms.txt` when it answers a question, and an [Ahrefs study](https://ahrefs.com/blog/llmstxt-study/) found most files were never requested. Treat it as cheap and harmless, not as a fix for visibility.

## What matters more

The engines do document these, and each can keep a site out of answers:

| Check | Why it matters |
| --- | --- |
| AI crawlers allowed in `robots.txt` | An engine does not cite a page its crawler may not fetch |
| No CDN or firewall block on AI crawlers | A block at the edge hides the site before `robots.txt` is read |
| Content in the HTML without JavaScript | Many crawlers do not run JavaScript |
| No `noindex` or `nosnippet` | Google documents that they keep a page out of its AI answers |

## Checking it with Manifold

[`aeo_get_site_readiness`](https://www.manifoldmcp.com/docs/tools/aeo#aeo_get_site_readiness) is free. It fetches your site the way the engines do and reports each check above with how much it matters, `llms.txt` included and marked as unproven.

## Questions

### Does ChatGPT read llms.txt?

OpenAI has not said that it does. Treat the file as a cheap extra and check the things the engines do document first: crawler access in robots.txt and at your CDN, and content in the HTML.

### Should I add an llms.txt file?

It costs little and does no harm, and agents that are pointed at it can use it. Do it after the fixes that are known to matter.

### How do I check my llms.txt?

Ask your agent to run Manifold's free readiness check on your domain. It reports whether llms.txt is served and follows the format, beside the checks that do block citations.

## Next

- [AI visibility MCP server](https://www.manifoldmcp.com/mcp/ai-visibility)
- [AEO tools in the docs](https://www.manifoldmcp.com/docs/tools/aeo)
- [OAI-SearchBot: how to allow it](https://www.manifoldmcp.com/blog/oai-searchbot)

## Sources

- [The /llms.txt file proposal (llmstxt.org)](https://llmstxt.org/)
- [Ahrefs study of llms.txt](https://ahrefs.com/blog/llmstxt-study/)

Published 13 October 2026. Updated 26 October 2026.