llms.txt is the emerging open standard for helping AI engines understand your website. This guide covers the official spec, where to host the file, how to write each section, and the two free tools that automate the entire workflow.
llms.txt is a Markdown file at the root of your domain (yoursite.com/llms.txt) that lists your most important URLs in a curated, AI-readable format. It was proposed by Jeremy Howard in late 2024 as an LLM-friendly companion to robots.txt and sitemap.xml. Anthropic, Stripe, Mintlify, and Vercel publish one. You should too β it takes 30 seconds with the right generator.
llms.txt is a single Markdown file served at the root of your domain (yoursite.com/llms.txt) that gives AI systems a curated map of your most important content. Unlike robots.txt (which controls access) or sitemap.xml (which lists every URL), llms.txt is selective and human-readable.
The standard was proposed by Jeremy Howard, co-founder of Answer.AI and fast.ai, in llmstxt.org in September 2024. The motivation: large language models like ChatGPT, Claude, and Perplexity have limited context windows. They cannot crawl an entire 10,000-page website to answer a single user question. They need a fast, curated entry point.
Think of llms.txt as your website's elevator pitch for AI engines. Where robots.txt says "yes" or "no", llms.txt says "here is what matters most".
Every spec-compliant llms.txt has the same five-part structure. Get any of these wrong and AI engines may parse your file incorrectly or skip it entirely.
A single # heading at the top of the file. This is the canonical name of your project, brand, or site.
# GrandRanker
A `>` blockquote immediately after the H1. One sentence describing what your site does.
> AI-powered SEO platform that automates the full content lifecycle.
Plain prose between the blockquote and the first H2. Use this for context, philosophy, or technical details.
GrandRanker handles the full content lifecycle from a single keyword.
Each section uses `##` and contains link items in the format `- [Title](URL): Description`. URLs must be absolute.
## Documentation - [Getting Started](https://grandranker.com/start): Quick setup
A special H2 named `## Optional`. AI engines treat these as lower priority than other sections.
## Optional - [Affiliate Program](https://grandranker.com/affiliate-program): 30% recurring
# GrandRanker > AI-powered SEO platform that automates keyword research, content creation, optimization, and publishing. GrandRanker handles the full content lifecycle from a single seed keyword: research, drafting, on-page optimization, internal linking, and one-click publishing to WordPress, Shopify, Webflow, Wix, and Framer. ## Documentation - [Getting Started](https://grandranker.com/start): Set up your first article - [Pricing](https://grandranker.com/pricing): Plans starting from $49/month - [Platform overview](https://grandranker.com/platform): CMS integrations ## Free Tools - [GEO Score Checker](https://grandranker.com/free-tools/geo-score-checker): Score any URL 0-100 - [llms.txt Validator](https://grandranker.com/free-tools/llms-txt-validator): Validate against the spec - [llms.txt Generator](https://grandranker.com/free-tools/llms-txt-generator): Auto-generate a file ## Optional - [Affiliate Program](https://grandranker.com/affiliate-program): 30% recurring commission
AI search adoption is no longer theoretical. ChatGPT processes over 800 million weekly users. Google AI Overviews appear on roughly 30% of US search queries. Perplexity, Gemini, and Claude all run their own crawlers and synthesize answers from web content.
These systems do not return ten blue links. They synthesize a single answer and cite three to five sources. Being one of those cited sources is the new "ranking on page one". And before you can be cited, you have to be efficiently discovered.
llms.txt solves the discovery problem at the structural level. Instead of asking AI engines to crawl your entire site to figure out what matters, you tell them directly. The file is small enough to fit inside any LLM context window and clear enough that downstream processing is reliable.
AI crawlers fetch llms.txt first and use it to prioritize which pages to fetch next. Skip the wait for full-site crawls.
Publishing the file signals to AI systems and the broader ecosystem that you actively support AI search.
Sitemap.xml lists everything. llms.txt lists what matters. Quality over quantity is what AI engines reward.
Costs nothing to publish. No SEO penalty. No conflict with robots.txt or sitemap.xml. Pure upside.
Adoption is accelerating among technically sophisticated companies. As of mid-2026, the following companies publish a public llms.txt:
The pattern is clear: technical platforms publish first because their users are most likely to be querying AI engines for documentation.
Two free tools that automate the whole workflow. Use the generator first to build the file, then the validator to confirm it meets the spec.
Crawls your homepage and sitemap, deduplicates and categorizes internal links by URL pattern (Documentation, Product, Blog, etc.), and outputs a spec-compliant Markdown file you can edit and download. From URL to ready-to-deploy llms.txt in 30 seconds.
Generate llms.txtValidates your llms.txt against the official spec. Runs 12 rule checks: H1 structure, summary blockquote, H2 sections, link format, absolute URLs, file encoding, line length, and more. Outputs a 0-100 score with line-numbered fixes and an embeddable verification badge.
Validate llms.txtIf you prefer to write the file by hand instead of using the generator, follow these six steps. The whole process takes about 15 minutes for a typical 50-100 page site.
Open your text editor. Create a new file named exactly llms.txt (lowercase, no extension changes). Save it with UTF-8 encoding.
First line: a single # followed by your brand or project name. Avoid emojis or special characters that might confuse Markdown parsers.
# Your Site Name
Leave a blank line, then add a > blockquote with a one-sentence summary. Aim for 80-160 characters that explain what your site does.
> A clear, concise description of your site or product.
Common sections: Documentation, Product, Free Tools, Pricing, Blog, Help & Support. Each H2 is "## Section Name" with link items underneath.
## Documentation - [Getting Started](https://yoursite.com/start): One-line description - [API Reference](https://yoursite.com/api): API documentation
Use the special "## Optional" section name for less-critical pages: legal pages, archive content, secondary tools. AI engines treat these with lower priority.
## Optional - [Privacy Policy](https://yoursite.com/privacy): Privacy statement - [Terms of Service](https://yoursite.com/terms): Legal terms
The file must be reachable at https://yoursite.com/llms.txt. WordPress: FTP into your site root. Webflow/Framer/Wix: use the static files or 404 redirect feature. Static hosts (Vercel, Netlify, Cloudflare Pages): place in public/ or static/ directory.
Most invalid llms.txt files share the same handful of errors. Avoid these.
Everything to know about llms.txt.
llms.txt is a Markdown file at the root of your domain (yoursite.com/llms.txt) that lists your most important URLs in a curated, AI-friendly format. The standard was proposed by Jeremy Howard (founder of Answer.AI and fast.ai) in late 2024. It is designed to help large language models like ChatGPT, Claude, and Perplexity understand your website structure without crawling every page.
GrandRanker writes AI-citation-ready content with proper structure, schema, and citations. From keyword to ranked article in 10 minutes.