llms.txt is a simple, human-readable Markdown file placed at the root of your domain (for example, example.com/llms.txt) that tells large language models what your site is about and where the content that matters actually lives. Think of it as a curated table of contents written for an AI, not a crawler directive.
Where robots.txt says what a bot may crawl and sitemap.xml lists every URL, llms.txt does something different: it hands the model a short, opinionated map of the pages you most want it to understand and cite.
Why llms.txt exists
Models have limited context and your site has a lot of noise — navigation, legal, duplicated boilerplate. llms.txt lets you cut through that and point the engine straight at the content that defines you: what you do, your docs, your pricing, your best explainers.
It is an emerging convention rather than a hard standard, but adoption is growing fast across docs sites, SaaS products and developer tools because it is cheap to add and genuinely clarifying.
The format
The structure is deliberately minimal: an H1 with your site or product name, an optional one-line summary as a blockquote, then H2 sections containing lists of links, each with a short note on what the page is.
A real example
# Acme Analytics
> Acme is a product analytics platform for subscription businesses.
## Docs
- [Getting started](/docs/start): install the SDK and send your first event
- [API reference](/docs/api): endpoints, auth, rate limits
## Product
- [Pricing](/pricing): plans, usage limits and add-ons
- [Security](/security): data handling, SOC 2, encryption
## Learn
- [What is product analytics?](/guides/what-is): the definitive explainerWhere to put it
- Serve it at the root: https://yourdomain.com/llms.txt (plain text, UTF-8).
- Keep an optional llms-full.txt for a longer, fuller dump if you have deep docs.
- Make sure it returns a 200 and is not behind auth, a redirect chain, or a bot block.
- Link only to pages your robots.txt actually allows AI crawlers to fetch.
Does it actually help?
Honestly: it is a signal, not a magic switch. Not every engine consumes llms.txt yet, and it will not rescue a site the crawlers cannot reach or that no trusted source ever mentions. But it is low-cost, it clarifies your most important pages, and it removes an easy reason for a model to misunderstand you. Pair it with answer-first content, schema and real third-party citations — that combination is what moves answers.
“llms.txt is the cheapest clarity you can give a model. It will not win the answer alone, but it removes a reason to lose it.”
Common mistakes
- Dumping every URL instead of curating the handful that define you.
- Linking to pages that robots.txt blocks for AI crawlers.
- Letting it 404 or sit behind a redirect after a site migration.
- Writing marketing fluff instead of plain, factual one-line descriptions.