Actually, Kage has two parts: a crawler that crawls pages and converts them to clean HTML by capturing the DOM after rendering in Chrome/Chromium, and a pack/serve component that packages the result as either a ZIM file for Kiwix or an executable file.
Related WHATWG discussion: https://github.com/whatwg/html/issues/3099