Free tool · No sign-up
llms.txt generator
Builds the file from your sitemap and your own page titles. And tells you plainly what it will and will not do, which most generators leave out.
Read this before you upload it
llms.txt will probably not help your AI search visibility
Roughly one domain in ten now serves an llms.txt, and close to 40% of those
files are empty plugin stubs. The reason for the enthusiasm is straightforward — it sounds
like the AI-era answer to robots.txt. The measurements do not support it.
Across a 90-day window covering more than 500 million AI bot visits, 408 requested the file. GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot and Google-Extended overwhelmingly ignore it and crawl your HTML directly. Google's 2026 AI optimisation guidance states outright that it is not needed for AI Overviews, AI Mode or any other generative Search feature. No major model provider — OpenAI, Anthropic, Google, Meta, Mistral — has committed to reading it.
Where it genuinely earns its place
As a routing map for AI coding agents. Cursor, Claude Code, GitHub Copilot, Cline, Windsurf and Aider all benefit from a clean index of a site's content, and llms.txt is a good way to give them one. If you publish developer documentation, an API reference or a large knowledge base, it is worth having.
If what you want is to be cited in AI answers, the file that decides that is your robots.txt — and the distinction between training crawlers and answer crawlers is where most sites get it wrong.
llms.txt · Questions
Straight answers about llms.txt
What is llms.txt?
A proposed standard: a markdown file at the root of your site that gives large language models a curated map of your important pages, so an AI reading your site does not have to work it out from navigation and HTML. The format is a heading with your site name, an optional summary, then sections of linked pages with short descriptions.
Does llms.txt improve my visibility in AI search?
On the current evidence, no. Across a 90-day window of more than 500 million AI bot visits, only 408 requested llms.txt — GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot and Google-Extended overwhelmingly skip it and crawl HTML directly. Google stated in 2026 that it is not needed for AI Overviews, AI Mode or any other generative Search feature, and no major provider has committed to using it as a signal.
Then why would I add one?
Because it does real work in a different place: as a routing map for AI coding agents such as Cursor, Claude Code, GitHub Copilot, Cline, Windsurf and Aider, and for any agent that needs a clean index of your content. If you publish developer documentation or an API, that is a genuine reason. If you are hoping for more citations in ChatGPT, it is not.
What actually decides whether AI systems can read my site?
robots.txt, which these crawlers do obey. The distinction that matters is between training crawlers and answer crawlers: blocking GPTBot keeps you out of training data but does not stop ChatGPT citing you, which is OAI-SearchBot. Our AI Crawler Access Checker reports both and explains the consequence of each.
How does this generator build the file?
It finds your sitemap through robots.txt or the usual paths, takes the shortest URLs — which are almost always the important ones — fetches each page for its real title and meta description, groups them by section, and formats the result. It uses what your own site already says rather than inventing summaries.
How many pages does it include?
Up to 30, chosen by shortest path. A complete map of a large site cannot be built inside a single request, and a truncated crawl would produce a worse file than the sitemap you already publish. For a bigger site, treat the output as a starting point and extend it by hand.
Where do I put the file?
At the root of your domain, served as plain text, so it resolves at yoursite.com/llms.txt. Check first whether you already have one — some platforms, Wix among them, generate it automatically, and the generated version may already point at capabilities you did not know you had.
Want to be found by AI properly?
Server-rendered content crawlers can actually read, structured data that describes what you do, and a deliberate decision about which AI systems may use it. That is the work. A text file at the root is not.