One of the earliest decisions in A11Y Cat was not glamorous at all. I did not want a tool that only worked in the perfect setup. That is why I kept both delivery paths alive: a bookmarklet for speed and a Chrome extension for the fuller product surface.
The bookmarklet matters because sometimes you are on a live page, under pressure, and you need to inspect structure, contrast, semantics, headings, links, forms, or modals without a whole install ritual. The extension matters because it gives me the richer interface, saved workflow, AI assist boundary, packaging, and runtime controls that a serious tool needs.
That split is not indecision. It is practical engineering. The repo documents both paths, the current status of each, and the reasons they exist. I would rather have two honest entry points than one supposedly elegant answer that fails the minute the context changes.
A lot of the project makes more sense once you understand that point. I was not building something to look tidy in a screenshot. I was building something I could actually use.
Documentation trail
README.mdextension/README.mdextension-guide.html
