The problem with most current Flash-to-HTML5 conversions is that they only barely support ActionScript, if at all. So anything meaningfully interactive (like minigames) is pretty much out.
Shumway makes it sound like "we got this" (which then discourages someone like me--someone who has a ton of knowledge of compiler design and a lot of background specifically with JavaScript-based language but almost no interest in duplicating effort, someone who would normally see this call to action and go "oh, I'll add this to my todo list"--from even spending much time researching the current space). Are you saying Shumway only "barely" supports ActionScript?
There are several such projects - swfdec, lightspark, gnash, etc. From what I've seen of them, they all underestimated the amount of effort. Most of them just petered out. So if you're going for it, I would worry less about duplicating effort and more about setting realistic goals for the project.