What llms.txt actually is, and is not, yet
llms.txt is a proposed convention from Answer.AI (Jeremy Howard), published in September 2024: a plain markdown file at your domain root that gives an AI system a curated map of your site, an H1 with your site or project name, an optional one-line summary, then H2 sections grouping links with short descriptions. It is deliberately simple, closer to a sitemap for language models than a technical standard.
Here is the part worth being upfront about: as of this writing, no major AI vendor, OpenAI, Anthropic, Google, or Perplexity, has published documentation confirming their crawlers or retrieval systems actually read llms.txt files. Google’s John Mueller said as much directly in June 2025. Publishing one will not hurt you, but treat it as a forward-looking bet, not a ranking lever with a confirmed payoff today. This tool checks your file against the community specification at llmstxt.org, that is the only thing we can actually verify.
Paste your llms.txt content above and click Validate to see the results here.
How we validate this
The specification we check against (from llmstxt.org): the file should start with a single H1 (#) naming the site or project, the very first line. An optional blockquote (>) directly after can give a one-line summary. Below that, H2 sections (##) group markdown links in the form - [Title](URL): optional description. A special ## Optional section is conventionally used for lower-priority links a crawler could skip under tight context limits.
What we flag: missing or duplicate H1s, content appearing before the H1, H2 sections with no links under them, and list items that are not valid markdown links. Headings deeper than H2 are not part of the spec and get a warning, not an error, since they will not break anything, they are just outside what the format defines.
What we cannot tell you: whether any AI system will actually fetch or use the file. As covered above, that is currently unconfirmed industry-wide, this tool only checks structural correctness against the published community spec.
Structural validation only, not a guarantee of AI crawler adoption. Specification source: llmstxt.org.