llms.txt Generator

What is llms.txt, and why generate one?

An llms.txt file is a small markdown document you place at the root of your website — reachable at https://yourdomain.com/llms.txt, exactly where you would put robots.txt. Its job is narrow and powerful: give large language models a clean, curated map of your site's most important content so they can understand it and cite it correctly. As AI assistants like ChatGPT, Claude, Gemini and Perplexity increasingly answer questions by reading the live web, the sites that hand these models a tidy summary win — they get represented accurately instead of being guessed at from a sprawling, ambiguous crawl.

The format, proposed at llmstxt.org, is deliberately simple. It opens with an H1 line naming the site, followed by a one-line blockquote summary (the > line) that captures what the site is in a single sentence. After that come one or more ## sections, each a bulleted list of markdown links with a short description per link. An optional ## Optional section signals links that an AI agent may safely skip when it is short on context budget. That is the whole spec — readable by humans and trivially parseable by machines.

How to use this llms.txt generator

The tool above turns a few inputs into a correctly structured file. Start by entering your site name and URL; the URL is used to expand any relative paths you list into absolute links, which is what AI crawlers prefer. Write a tight one-line summary — this is the single most important line in the whole file, because it is the first thing a model reads and the sentence it is most likely to paraphrase. Add an optional details paragraph if you want to give extra context such as pricing, regions, or available SDKs.

Next, list your key pages, one per line, using the simple path | Title | description format. Lead with the pages that define your site: documentation, an API reference, a pricing page, your best evergreen articles. Keep descriptions short and factual — you are writing for a model that will quote you, not for marketing copy. Use the Optional pages box for lower-priority links like terms and privacy. Hit Generate, then Copy or Download, and upload the file to your domain root. That is it.

llms.txt vs robots.txt vs sitemap.xml

These three files do different jobs and you should publish all of them. Your XML sitemap is an exhaustive machine list of every URL you want indexed — comprehensive but undifferentiated. Your robots.txt is an access-control list: it tells crawlers, including AI crawlers like GPTBot and ClaudeBot, which paths they may fetch at all. llms.txt is neither of those: it is an editorial summary, a short hand-picked guide to what matters most, written specifically so a language model can orient itself quickly. Think of the sitemap as the full index, robots.txt as the door policy, and llms.txt as the welcome desk that points visitors to the right floor.

Tips for a high-quality llms.txt

• Lead with clarity. The blockquote summary should answer "what is this and who is it for?" in one sentence. Avoid jargon and superlatives.
• Curate, do not dump. Ten well-chosen links beat two hundred. The point is to surface the canonical, high-value pages, not mirror your sitemap.
• Write factual descriptions. Each link's description should state what the page actually contains. Models reuse this text.
• Use absolute URLs. They remove ambiguity for crawlers that may encounter the file out of context. This generator builds them for you from your site URL.
• Keep it current. Regenerate when your information architecture changes — a stale llms.txt pointing at dead pages hurts more than none.
• Consider an llms-full.txt. Larger documentation sites sometimes publish a second, expanded file with full page content inlined. Start with llms.txt; add the full version if your docs warrant it.

Once your llms.txt is live, pair it with a tuned AI robots.txt to control which crawlers may access your content, and use our token counter if you are also building LLM features and need to budget prompts. All three tools are free, run entirely client-side, and never upload your data.