You can find why it was complicated [0] and why it took so long [1] on GitHub. I do wish I had been faster on it. That said, the RCE required the user to be using a vibe-coded web-based chat overlay that didn't properly sanitize chatter inputs, so has a (hopefully) limited scope.
Thanks for chiming in and linking that, that was an interesting read. As much as I understand CEF's reasoning for making this shift, it's baffling that they don't provide a universal way to implement what you now had to implement yourself.