🤖 llms.txt Generator & Validator

/llms.txt is an emerging spec by Jeremy Howard (Answer.AI) — a clean, structured Markdown file that tells large language models which parts of your site to read and how to navigate them. Think of it as a sitemap optimised for AI rather than search bots.

The spec lives at llmstxt.org. It's not yet a W3C standard, but Anthropic, Perplexity, Cursor and several IDE assistants already read it where available.

Generate llms.txt

Primary docs (the „must-read" pages)

One line each — [Title](URL): one-sentence description. Tip: paste your top 5–10 evergreen pages.

Optional resources (the „nice-to-have")

Output (paste at https://yoursite.com/llms.txt)

What goes in llms.txt?

The spec defines a Markdown document at the root of your site with this structure:

# Site title              ← required, one H1

> A one-line blockquote summary  ← optional but recommended

Optional paragraph of detail.   ← any number of paragraphs

## Section title                ← any H2 grouping
- [Doc title](URL): description ← list items, one resource each

## Optional                     ← reserved keyword: nice-to-have, not required for LLMs

Why bother? LLMs and AI agents that respect llms.txt use it to prioritise which pages to read and in what order. Without one, they crawl HTML pages and try to figure it out — leading to misquotes and missed coverage.

Many sites pair llms.txt with parallel Markdown copies of pages (e.g., /api.md alongside /api). The Markdown copy is what the LLM actually reads.