llms.txt
A homepage for language models.
Search engines have sitemap.xml. AI agents have llms.txt. Proposed by Jeremy Howard in 2024, it's a simple Markdown file at /llms.txt that gives LLMs a curated tour of your site — without forcing them to crawl every page and render JavaScript.
Why it matters
- LLMs have tiny context windows compared to the modern web.
- Crawling JS-heavy sites is slow and unreliable for agents.
- You decide what content represents your product, not a scraper.
- Major scanners (including this one) explicitly check for it.
Minimal valid example
# Acme Inc.
> Acme builds rocket-propelled widgets for the modern road runner.
## Docs
- [Getting Started](https://acme.com/docs/start): Install in 60 seconds.
- [API Reference](https://acme.com/docs/api): Every endpoint, with examples.
## Optional
- [Changelog](https://acme.com/changelog): What shipped this week.
How it works
Place the file at https://yourdomain.com/llms.txt. The format is strict: a single H1 (your name), an optional blockquote summary, free-form paragraphs, then H2 sections containing markdown link lists in the form - [Title](url): description.
Optional companion: llms-full.txt
A second file, /llms-full.txt, embeds the entire content of the URLs you reference so an agent can ingest your knowledge base in a single fetch. Most projects only ship llms.txt and reach for llms-full.txt when their docs are stable.
Generate a starter file with our llms.txt generator, then commit it next to robots.txt.