llms.txt: What It Is and Whether It Actually Does Anything in 2026

What llms.txt actually is

llms.txt is a proposed convention: a markdown file served at yourdomain.com/llms.txt that gives large language models a curated map of your site — the pages that matter, in plain text, with one-line descriptions. It is a suggestion to AI systems about what to read, not an instruction they are obliged to follow.

The format is deliberately simple. An H1 with your site name, an optional blockquote summary, then H2 sections containing markdown links with short descriptions. Some sites also publish llms-full.txt, which inlines the actual content instead of linking to it.

Who actually reads llms.txt in 2026

This is the part most guides skip. The honest position of each major AI company, as publicly stated:

A useful comparison: John Mueller at Google likened llms.txt to the keywords meta tag — a self-declared hint that search engines eventually learned to ignore because publishers had every incentive to overstate. That is a fair worry, and it is the reason nobody should build an AI strategy on this file.

What llms.txt is genuinely good for

1. Agents and developer tooling

Coding agents, documentation assistants, and MCP-style integrations increasingly look for a plain-text entry point. If you publish an API or technical docs, llms.txt gives an agent a clean starting index instead of making it guess from your navigation.

2. Forcing you to decide what matters

Writing the file means picking the twenty pages that represent your business and describing each in one line. Teams routinely discover during this exercise that their most important service page has no clear description anywhere on the site. The document is cheap; the editorial decision behind it is the real deliverable.

3. Perplexity prioritisation

Of the consumer AI engines, Perplexity is the one that has said it uses the file. If Perplexity is a meaningful referral source in your category, this is a real if modest reason to ship one.

4. Cheap insurance if the convention gains traction

The file takes about an hour to write and costs nothing to host. If adoption grows, you are already covered. That is a defensible reason to do it — but it is a very different argument from “this will improve your AI rankings”, and you should not let a vendor sell you the second one.

llms.txt vs robots.txt vs sitemap.xml vs schema markup

If you only have budget for one of these, it is not llms.txt. Getting crawler access right comes first — that is covered in can ChatGPT see my website — and structured data second.

How to ship an llms.txt correctly

Step 1 — Pick the pages, not all the pages

• Choose 15–30 URLs: core service pages, pricing, documentation, and your strongest explanatory articles.
• Write a one-line description for each that states what the page answers, not what it is called.
• Leave out tag archives, thin category pages, and anything you would not want quoted back at you.

Step 2 — Write the file in markdown

• Start with # Your Company Name, then a one-paragraph blockquote describing what you do.
• Group links under H2 headings such as ## Services and ## Guides.
• Each entry is a markdown link followed by a colon and the description.

Step 3 — Serve it as plain text with a 200 status

• Upload it to the web root so it resolves at yourdomain.com/llms.txt.
• Serve it as text/plain, not as an HTML page wrapped in your theme.
• Verify with curl -I https://yourdomain.com/llms.txt and confirm the first line says 200.

Caution — the failure mode we see most often: if you generate llms.txt from a CMS route or plugin rather than a static file, the body can render perfectly in a browser while the server still returns 404 in the HTTP header. The page looks live to you and is invisible to every crawler. In WordPress this happens when a custom rewrite outputs content without calling status_header(200). Always check the header with curl -I, never the browser window.

Should you bother?

1. Have you fixed crawler access and schema first?

If OAI-SearchBot cannot reach your pages, or your site has no structured data, llms.txt is decoration on a locked door. Do those two first, in that order.

2. Do agents or developers consume your content?

Documentation sites, APIs, and technical products get the most from it, because that is where agent traffic is real today. A local restaurant gets close to nothing.

3. Is Perplexity worth anything in your category?

Check your referral traffic. If Perplexity sends measurable visits, the file has a concrete rationale rather than a speculative one.

4. Can you keep it current?

A stale llms.txt pointing at deleted pages is worse than none, because it advertises neglect to the exact systems you are courting. Only ship it if it will be updated when the site changes.

CiteLogics — spend your effort where it actually moves AI visibility

The AI Visibility Audit scores five dimensions — crawlability, content understandability, AI citation presence, technical performance, and AI Overview presence — so you can see whether llms.txt belongs anywhere near the top of your list. Usually it does not.

• AI Visibility Audit — full AIVS Score report + 30-minute call. THB 2,000, delivered in 24 hours.
• Big Win (Audit + WebFix) — audit plus full implementation, schema and rendering included. THB 20,000.

Get your free AI visibility scan, or contact sales@citelogics.com.

Frequently asked questions