# llms.txt — from curiosity to working convention

*2026-02-02 · Discoverability*

> The llms.txt file has moved from novelty to a low-cost convention worth adopting, even as honouring it remains uneven across crawlers.

**What changed.** The **llms.txt** convention — a root-level Markdown file that tells AI
crawlers what a site is about and which pages are authoritative — has steadily moved from
curiosity to common practice across documentation sites, tools, and content publishers.

**Why it matters.** It remains a *convention*, not a universally honoured standard: not
every crawler reads it today. But the economics are lopsided. It costs almost nothing to
publish, it can't hurt, and it positions you correctly for where AI-mediated discovery is
heading. Pairing it with a [clean machine-readable substrate](/concept) is the point.

**What to do.** If you haven't already, publish an [llms.txt](/guides/what-is-llms-txt)
listing your most important pages with one-line descriptions, reference it from
robots.txt, and add a fuller llms-full.txt if your library justifies it. This very site
serves both — verify them at [/llms.txt](/llms.txt).

*Tracking: llms.txt, llms-full.txt, AI crawler directives.*