What Is llms.txt?

llms.txt is a plain-text file at a website's root that lists its key pages for AI models to read directly, instead of parsing full HTML. Jeremy Howard of Answer.AI proposed the convention on 3 September 2024. It borrows robots.txt's root-path convention but does a different job: guiding AI assistants like ChatGPT and Perplexity to readable content rather than controlling crawler access. It is a community proposal that no standards body or AI lab has formally adopted.

What goes inside an llms.txt file?

An llms.txt file has four elements, in order, and only the first is required. It opens with an H1 site name, the one part the spec mandates, then an optional one-line blockquote summary. After that can come any number of markdown ## sections that group links to key pages, each with a short description.

The format is deliberately plain. It is human-readable and machine-readable at once, served at /llms.txt the same way /robots.txt is.

A model, or a person, can fetch it in one request with no login and no JavaScript.

Some sites also publish a companion llms-full.txt. That file inlines the full text of every linked page, for models that can accept one large fetch instead of following each link separately.

How is llms.txt different from sitemap.xml and robots.txt?

llms.txt, sitemap.xml, and robots.txt all sit at a site's root, but they serve three different readers. The table below sets them side by side.

	llms.txt	sitemap.xml	robots.txt
Audience	AI models and assistants	Search engine crawlers	All crawlers
Format	Markdown, human- and machine-readable	XML	Plain-text rules
Purpose	Curated list of key pages	Complete list of URLs to index	Allow/disallow crawl rules
Official status	Community proposal	Backed by all major search engines	Formal standard, RFC 9309

The practical difference is scope. A sitemap lists everything you want indexed. An llms.txt file curates only the pages you most want an AI model to read, in a format it can lift without parsing your HTML first.

Does llms.txt actually affect AI citations?

llms.txt has no confirmed effect on AI citations. No major AI lab has published whether its crawlers or retrieval systems consult the file at all. Treat any claim of a guaranteed AI benefit from having one as unverified.

What the file certainly does is remove ambiguity for any system that does read it. It hands the model an explicit, curated list instead of forcing it to infer structure from HTML. Once it is generated automatically, it costs nothing to keep.

That is the honest case for having one: a low-effort complement to the same answer-first structure that AEO and GEO already ask for, on the crawlable foundation ordinary SEO needs anyway.

What does our own llms.txt contain?

This site's llms.txt is generated at build time by a route handler, not written by hand. It reads from the same wiki index every other page reads, so the file cannot list a draft article or miss a newly published one.

It opens with the site name and a one-line summary, then lists each service page, the company pages, and every published Digital Hub article grouped by category. Each line carries its own short description, pulled from the same data the page itself renders from.

The reason we build it this way is drift. A hand-edited or plugin-generated llms.txt is written once and rarely touched again, so it slowly stops matching the site. A file generated from the live index updates itself on every publish, which is the one maintenance failure a static file cannot avoid.

Frequently asked questions

Is llms.txt an official web standard?

No. llms.txt is a community proposal, first published by Jeremy Howard of Answer.AI on 3 September 2024, with its spec at llmstxt.org. It has no ratification from a standards body, and no search engine or AI lab has adopted it as an official requirement. Contrast that with robots.txt, which is a formal standard (RFC 9309, linked in the table above). Publishing an llms.txt file is optional and carries no penalty either way.

Do AI models actually read llms.txt?

Unconfirmed. No major AI lab has stated whether its models read or prioritize the file, so nobody can honestly promise it changes how you are cited. Publisher-side adoption is real: technical companies publish the file, but adoption on the publishing side says nothing about whether any model actually reads it. The safe framing is that llms.txt helps any system that chooses to read it and hurts nothing if none do.

What is the difference between llms.txt and llms-full.txt?

llms.txt lists your key pages as links with short descriptions; llms-full.txt inlines the entire text of those pages into one file. The short version suits models that follow links. The full version suits models that prefer a single large fetch. Neither is more official than the other, and adoption of both has not been measured publicly by any engine.

Where do I put the llms.txt file?

At the root of your domain, reachable at yoursite.com/llms.txt. It must return as plain text with no login and no JavaScript required, exactly like robots.txt. If it sits in a subfolder or needs a script to render, the tools looking for it will not find it.

Does my small business website need one?

Not strictly. An llms.txt file is a low-effort extra, not a foundation, so it belongs after the basics: a crawlable, well-structured site with answer-first pages. If your site is generated by a modern framework, adding a build-time llms.txt is close to free. If it means hand-maintaining a file that will drift, the effort rarely earns its keep for a small site.

How do I check whether a site already has one?

Request /llms.txt on the domain, either by typing the path into a browser or running curl against it. A compliant file returns as plain text immediately. If you get a 404 or an HTML page instead, the site has no llms.txt at that address.

Where this fits your site

Storming Solutions builds and optimizes Malaysian websites to be read cleanly by both Google and AI models, and a build-time llms.txt is one small part of that. Nobody can honestly sell the file as an AI-ranking guarantee today, so on the sites we build it simply comes included, as free housekeeping.

Want to know whether AI systems can actually read your site the way this one is built to be read? Ask us for a free AI Visibility Report over WhatsApp, or see how we run the work on our SEO, AEO & GEO service page.