Add llms.txt to Your Site

Add an llms.txt file at your site root that lists your most important pages as a clean, linkable index agents can read before they crawl your site.

What is llms.txt?

llms.txt is a Markdown file at the root path /llms.txt that gives AI agents a curated, linkable index of a site's most important pages. Jeremy Howard of Answer.AI proposed it on 3 September 2024 (Answer.AI, 2024); llmstxt.org maintains the specification. The format is ordinary Markdown: an H1 with the site name is the only required section, followed by an optional blockquote summary and H2 sections of - [Title](url): one-line note links; a section named "Optional" marks links an agent may skip when context is short (llmstxt.org, 2026).

Why does llms.txt matter for agents?

llms.txt matters because it collapses discovery into one small fetch: an agent with a limited context budget reads a single file and learns which pages answer which questions, in your own words, without parsing layout HTML — the Layer-1 declaration of the discoverability dimension. Consumption is unproven: in June 2025 John Mueller of Google wrote on Bluesky that no AI system currently uses llms.txt (Search Engine Roundtable, 2025). Publication keeps growing: llmstxt.org lists OpenAI, Anthropic and Gemini serving the file for their own developer docs, and Chrome's Lighthouse audits for it in its agentic-browsing category (Google, 2026).

How do you implement llms.txt?

Implement llms.txt in three moves: write the file, serve it at the root, confirm it resolves.

1. Write a Markdown file: an H1 with the site name, an optional > blockquote summary, then ## sections (Docs, Products, Reference) of - [Title](https://example.com/page): one-line description links, with skippable links under ## Optional.
2. Serve it at the exact path https://yoursite/llms.txt — root, no redirect chain — as text/plain or text/markdown.
3. Verify: curl -sI https://yoursite/llms.txt returns 200, and the body starts with a Markdown heading whose links resolve.

The Agent-Readiness Audit's llms-txt check (8 of 100 points) passes when /llms.txt at the domain root returns 200 and starts with a Markdown heading; a companion /llms-full.txt that inlines the full text earns the separate llms-full check (4 points). This site serves both files as live proof.

llms.txt — frequently asked questions

Does any AI system read llms.txt?

Not provably, as of June 2025: John Mueller of Google wrote on Bluesky that no AI system currently uses llms.txt (Search Engine Roundtable, 2025). Publish it as a low-cost discovery aid and keep robots.txt, the sitemap and internal links as the primary signals.

Is llms-full.txt part of the llms.txt specification?

No. llmstxt.org specifies only llms.txt; llms-full.txt, which inlines the full text of the linked pages, is a community convention. This site serves both files, and the Agent-Readiness Audit scores llms-full.txt as its own check worth 4 of 100 points.

Where must llms.txt be served?

At the root path /llms.txt, or at a subpath such as /docs/llms.txt for a sub-site (llmstxt.org, 2026). The Agent-Readiness Audit fetches the root path only, so publish it there first.

Related: how llms.txt differs from robots.txt and agents.txt · the llms.txt spec entry · the discoverability dimension · the Agentic Web Lexicon · audit your site

← Agent-Readiness · .md