llms.txt — store profile for AI agents

More and more shoppers ask an assistant rather than a search engine — ChatGPT, Perplexity, Gemini. An assistant sees a store differently from a person: it needs a short machine-readable description — an llms.txt file, not a banner on the homepage.

The llms.txt module generates a/llms.txtfile to the open llmstxt.org standard — a curated document in which you tell AI agents what this store is.

What goes into the file

• Store name and description — short, in your own words.
• A link to the read-only MCP catalogue — a typed interface through which an agent reads products directly instead of guessing them from markup.
• Top categories.
• Curated SEO facet pages — the selections you consider your shop window.
• Legal pages.

The operative word is curated. This is not a catalogue dump: the full page inventory is already sitemap's job. Here you decide what to show rather than handing a crawler the whole site at random.

How it works technically

The file is static: a daily cron job regenerateswww/llms.txtand the web server delivers it directly — zero PHP per public request. The module has no admin page of its own; its settings live in the shared plugin form, as with Sitemap.

What Pro adds

llms.txt Pro provides per-category feeds instead of a single combined file, plus an access log of AI agent requests — so you can see who reads what, rather than guessing.

Honestly, about expectations

This is a young standard, and no vendor can guarantee that a particular assistant will read it or recommend you. We do not guarantee it either — anyone who does is selling air. The argument is simpler: the file costs nothing once installed, and the alternative is hoping an agent works your catalogue out on its own.

Nearby: AI modules and SEO. The base tier ships inside the Community build.