What this check tests
The audit requests /llms.txt at your site root. The file counts as found when the server answers 200 and the response is not an HTML page. The audit rejects a 200 as a soft 404 only when its media type is text/html and the body is empty or starts with < (after optional whitespace and a byte-order mark); the same body under another media type, such as text/plain, is accepted.
If the root file is missing, the audit also tries llms.txt on the docs., developers. and developer. subdomains of your domain. The result is pass or fail; there is no partial. The proposal also allows an llms.txt under a subpath such as /docs/llms.txt, but this check does not look there.
Why it matters to agents
The proposal describes llms.txt as a curated, markdown index that an agent reads on demand to find the right pages, and says the authors expected inference-time use more than training. In practice that means coding assistants and tools that fetch documentation.
Be realistic about reach. Google's documentation on AI features says you do not need new machine-readable files or AI text files to appear in AI Overviews or AI Mode. Treat llms.txt as worth having for coding agents and tools that fetch it, not as a lever for visibility in Google or chat assistants. See the llms.txt validator page for the same caveat with more sources.
How to fix it
Create a plain-text markdown file and serve it at https://yoursite.com/llms.txt with a 200 status and a text content type (text/plain or text/markdown).
# Example Co
> Example Co sells project-planning software. This file lists the pages that best explain the product and its API.
## Docs
- [Getting started](https://example.com/docs/start): Install and first project
- [API reference](https://example.com/docs/api): Endpoints, auth and limits
## Policies
- [Pricing](https://example.com/pricing): Plans and billing
- Put it in your static or public folder so the web server returns it as a file, not through a catch-all route that renders your 404 page as HTML.
- List the pages you would send a new engineer to, not every URL. A sitemap already covers the full list.
- After deploying, request the URL and confirm the
Content-Typeheader and the status code.
Once the file exists, llms.txt follows the format and llms.txt links resolve start scoring it. Background and platform notes are in what llms.txt is; setup pages exist for WordPress, Next.js, Shopify, Docusaurus, Nuxt and Astro.