I've had the same idea. Operationally, what's needed is speed.

I do not care to learn some obscure, poorly documented API by trial and error. If I could write a failing test or three in a playground then post it in an appropriate chat window and have two or three people claim it at once, with the winner earning 50% of the prize, I would utilize this with some frequency.

The problem with an SO-style solution is that I don't get confirmation that anyone is working on it. So if my project is under a stiff deadline then I have to keep trying myself.

That said, yes, someone please build this. There is a market for people like me that have wasted days on ffmpeg just to get X to do a Y without a Z.