Turn a website into an AI-readable knowledge bundle.

Paste a URL below and get back an Open Knowledge Format bundle: a zip of clean, linked markdown files that AI agents can read instead of scraping HTML.

• Open index.md first. It lists every page with a one-line description, so an agent (or you) can see how the site is laid out before opening every file.
• Drop the unzipped folder into a coding agent or chat tool's project files so it can read the whole site as clean markdown instead of scraping HTML.
• Use it for RAG. The files are already split one-per-page and cross-linked, so you can chunk and embed them directly instead of scraping raw HTML yourself.
• Host it at /okf/ on your own site if you want AI agents to discover it. That's what this site itself does with its own bundle.

• No JavaScript rendering. The crawler reads only the initial server-rendered HTML, so pages that need client-side JS to show their main content won't extract well.
• Capped at 30 pages and about 75 seconds of crawl time per site. Larger sites get a partial bundle, and index.md notes when that happens.
• Respects robots.txt. The crawler skips disallowed pages, and refuses any site publishing a Content-Signal: ai-input=no header.
• Rate-limited to a few generations per hour per visitor, since this is a shared tool running on a small server.
• Content extraction is best-effort. The main-content heuristic can grab too much or too little on unusually structured pages.