llms.txt

llms.txt is a proposed web standard: a single Markdown file at the root of a website that gives large language models a curated map of what the site is, what it offers, and which pages matter — the AI-era counterpart to robots.txt.

What is llms.txt?

llms.txt is a plain-text convention for making a website legible to AI systems at inference time. Instead of forcing a model to crawl HTML, strip navigation, and guess which pages are authoritative, the site owner publishes one Markdown file — /llms.txt — that states the site's name, summarizes its purpose, and lists curated links to the content a model should actually use.

The format is deliberately simple. A compliant file starts with an H1 naming the site or product, followed by a blockquote that summarizes it in a sentence or two. After that come H2-headed sections of bullet links — documentation, pricing, API references, policies — each with a short description. A special "Optional" section marks links a model may skip when its context window is tight. A companion file, /llms-full.txt, inlines the full text of key documents so a tool can ingest everything in a single request.

Why does llms.txt matter for AI search?

Because AI answer engines quote what they can cleanly read, and most websites are not cleanly readable. Websites were designed for browsers: heavy markup, JavaScript rendering, cookie banners, and navigation chrome surround the actual content. When an AI assistant answers a question by browsing the web, it burns most of its context window on boilerplate — or fails to render the page at all.

llms.txt flips the economics. The site owner, who knows which pages are current and authoritative, hands the model a pre-digested map. The result is more accurate citations, fewer hallucinated product details, and a better chance that an answer engine describes your offering the way you would describe it yourself. For companies whose buyers increasingly research through ChatGPT, Claude, Perplexity, or Gemini, that is discoverability infrastructure — the same role SEO markup played for Google.

How does llms.txt compare to robots.txt, sitemaps, and structured data?

These files coexist; they answer different questions at different points in the pipeline.

The distinction that matters: sitemaps and robots.txt are exhaustive and mechanical, while llms.txt is curated and editorial. A sitemap says "everything"; llms.txt says "start here, this is what's true about us."

What addllms.com is building

addllms.com is an early-stage property in build by ROIZILLA, LLC. The planned capability is llms.txt tooling: generating a compliant file from an existing site, validating structure against the convention, and keeping the file in sync as content changes. That tooling is not live today. This page is an educational reference on the standard itself — no products, services, or customer features are offered yet.

Frequently asked questions

Is llms.txt an official web standard?

Not yet. It is a proposed convention introduced in 2024, not an IETF or W3C standard. Adoption is voluntary and growing among documentation sites, developer tools, and AI-native companies.

Does llms.txt replace robots.txt?

No. robots.txt controls what crawlers may fetch; llms.txt tells a model that is already using your site which content is worth reading. Most sites should publish both.

Where does the file go?

At the domain root, served as /llms.txt. The convention also defines /llms-full.txt for tools that want the full content in one fetch.

Do AI models actually read llms.txt today?

Some inference-time retrieval tools and coding agents fetch it when available, and several AI platforms honor it as a hint. Treat it as a low-cost, high-upside signal while answer-engine adoption matures.

What should a good llms.txt contain?

An H1 with the site name, a blockquote summary, short context paragraphs, and H2-sectioned curated link lists with one-line descriptions — plus an Optional section for skippable material.

Who operates addllms.com?

addllms.com is owned and operated by ROIZILLA, LLC. It is a property in build: llms.txt generation and validation tooling is planned, and this page currently serves as an educational reference.