Tools like ChatGPT, Claude, and Perplexity want to understand what your site offers before they cite it. The llms.txt standard delivers that context directly to the machine.
Keysonar's generator goes further than static templates: it tests whether GPTBot, ClaudeBot, PerplexityBot, and Google-Extended can actually reach your pages, analyzes your content categories, and flags issues to fix. The output is grounded in your site's real content.
No credit card. 3 free domains per day on the free tier.
llms.txt is a Markdown file at the root of a website (e.g., example.com/llms.txt) that gives AI search engines a structured map of the site: what it does, what content matters, and how each section is organized.
The format is defined by the llmstxt.org informal spec and is read by ChatGPT, Claude, Perplexity, and other LLMs to improve citation accuracy. Without it, AI engines must crawl every page and guess your site's structure — often citing stale or wrong information.
Under a minute, end to end. No setup, no API keys, no client integration.
Type or paste the site. Bare hosts (example.com) and full URLs both work.
robots.txt + sitemap.xml → 25-30 representative URLs analyzed for content type and SEO signals.
We check whether GPTBot, ClaudeBot, PerplexityBot, and Google-Extended can actually access your site.
Review validation in 3 categories, then download and upload to your site root.
Most generators produce a static template. This one is grounded in your site's actual content and AI bot accessibility.
A perfect llms.txt is useless if your robots.txt blocks GPTBot. We probe all four major AI engines (GPTBot, ClaudeBot, PerplexityBot, Google-Extended) in real time and tell you exactly which are reachable.
We classify pages by content type (homepage, category, product, blog, FAQ, about, contact) from URL patterns and on-page signals, so your manifest groups links the way AI engines expect to read them.
Errors (spec violations), Warnings (size/link concerns), and GEO Suggestions (citation-quality improvements: short headings, missing descriptions, slug-only link text). Most validators stop at error/warning.
We crawl your live site — homepage, categories, products, blog — and build the manifest from what's actually there. No generic templates, no manual editing required for a working baseline.
Quick comparison against the two most-used alternatives.
| Feature | Keysonar | llmstxt.org | Firstbatch |
|---|---|---|---|
| Manifest generation | |||
| Live AI bot accessibility probe (4 engines) | |||
| GPTBot robots.txt check | |||
| Content category inference | partial | ||
| Spec validator with error/warning split | |||
| GEO citation-quality suggestions | |||
| Free to use |
If you want AI search engines to cite your site accurately, yes. Without llms.txt, AI engines must crawl every page and guess your site's structure, which often leads to outdated or wrong citations.
ChatGPT's browse mode and Perplexity actively use llms.txt as a citation source. Anthropic's ClaudeBot fetches it for retrieval-augmented responses. Google has not officially committed to the standard, but Google-Extended respects it as a robots.txt-like signal.
sitemap.xml lists every URL on your site for search engine indexing — a complete inventory. llms.txt is a curated narrative: what your site is about, which sections matter, and contextual descriptions that help AI engines understand and cite your content correctly.
Yes. Sign up for a free Keysonar account and you can generate llms.txt files for up to 3 different domains per day. No credit card required, no trial period.
Yes — the 'For a different site' option is designed for exactly this case. Agencies and freelancers can analyze any client domain without connecting it via API.
Sign up free and generate your first llms.txt in under a minute.