llms.txt is a proposed standard, published at llmstxt.org in 2024, for a Markdown file placed at /llms.txt that helps large language models make sense of a website. It contains a title, a short summary and lists of links to the pages that matter most, ideally pointing to clean Markdown versions. It is a convention, not an enforced protocol, and as of late 2026 no major AI search provider has documented that it uses llms.txt to choose sources or rank answers.
What the file looks like
The proposal sets out a simple structure: an H1 with the project or site name, a blockquote summary, optional free text, then H2 sections containing Markdown link lists, each link followed by an optional note. A section headed “Optional” marks links that can be skipped when context is short.
# Quillbrook
> Quillbrook is accounting software for small agencies. It handles invoicing, expenses and time tracking, with integrations for Stripe, QuickBooks and Xero.
## Product
- [Pricing](https://ledgerly.example/pricing.md): Plans, limits and billing terms
- [Integrations](https://ledgerly.example/integrations.md): Supported tools and setup
## Docs
- [API reference](https://ledgerly.example/docs/api.md): REST endpoints and authentication
## Optional
- [Changelog](https://ledgerly.example/changelog.md)llms.txt compared with robots.txt and sitemaps
robots.txt tells crawlers what they may fetch. An XML sitemap lists URLs for search engines to discover. llms.txt does neither: it is a curated reading guide meant for use at the moment an AI tool or agent is trying to understand the site. Adding it does not grant or block access to anything.
The proposal also suggests an llms-full.txt variant that contains the full text of key pages in one file, and Markdown versions of pages at the same URL with .md appended.
Should a SaaS company add one?
Cost is low, so many developer-tool and API companies publish one, mostly for coding assistants and agents that read documentation. For marketing pages, the benefit is uncertain because the main AI search crawlers have not said they read it. Google representatives have said publicly that Google Search does not use it.
A sensible position: publish one if your docs are a big part of how customers adopt the product, keep it accurate, and do not expect it to change AI search visibility on its own.
Common mistakes
Expecting it to work like robots.txt
llms.txt does not control crawling or training. Use robots.txt for access rules.
Letting it go stale
A file that links to old pricing or removed pages gives any tool that does read it wrong facts. Generate it from the same source as your sitemap if you can.
Questions
Neither OpenAI nor Google has documented llms.txt as an input to its search products as of late 2026. Some coding tools and agent frameworks do read it when pointed at a site.
At the root of the site, at /llms.txt, served as plain text or Markdown. The proposal also allows it in a subpath.
There is no known downside for search. It is just another file on your server.
