Updated 2026-07-23 · 8 min read
An llms.txt file is a simple Markdown file at the root of your site that gives AI tools a curated map of your most important content. Think of it as a friendly index for language models: not a ranking switch, not a firewall, but a way to say "here is what matters and where to find it." Used correctly it is a small, useful layer on top of a well-built site.
It is a human- and machine-readable directory of key pages, written in Markdown, served at /llms.txt. It is not a replacement for crawlable HTML, internal links or structured data, and it does not force any model to read it or to rank you higher. It is documentation, adopted voluntarily by the tools that choose to consume it.
The convention is deliberately minimal: an H1 with the site name, an optional blockquote summary, then Markdown link lists grouped under H2 sections. Each link is a URL with a short description.
# Norrly
> Measured design breakdowns and vibe-coding prompts from hundreds of top websites.
## Guides
- [What is vibe coding](https://norrly.io/guides/what-is-vibe-coding): A practical introduction.
- [Generative Engine Optimization](https://norrly.io/guides/generative-engine-optimization): How to get cited by AI.
## Core pages
- [Inspiration library](https://norrly.io/inspiration): Browse measured design breakdowns.
- [State of Web Design](https://norrly.io/state-of-web-design): Our annual data report.
Some sites also publish llms-full.txt, which inlines the full text of key documents rather than just linking to them, so a model can ingest everything in one request. This is useful for documentation-heavy products, but it duplicates content and must be kept in sync. For most sites, a curated llms.txt with clean links is enough.
Treat it as the smallest step in a broader GEO strategy, layered on top of the fundamentals: crawlable pages, clear entities, connected JSON-LD and original content worth citing. A perfect llms.txt on a thin, client-only site achieves nothing; a solid site gains a modest, low-cost convenience.
There is no evidence that llms.txt directly improves Google rankings, and major search engines have not confirmed using it as a ranking input. Its value is as an AI-readable directory that a subset of tools may consult. Publish it because it is cheap and tidy — not because it is a growth hack.
It is a Markdown file at your site root (/llms.txt) that gives AI tools a curated, human-readable map of your most important pages with short descriptions. It is documentation for language models, not a ranking mechanism.
There is no evidence it directly improves Google rankings, and search engines have not confirmed it as a ranking factor. It can make your key content easier for some AI tools to discover, but it does not replace crawlable HTML, internal links or structured data.
llms.txt is a curated index of links with descriptions; llms-full.txt inlines the full text of key documents in one file so a model can ingest everything at once. Full versions suit documentation-heavy products but must be kept in sync with the source content.