Looks good.

And I agree about shoulders being the most difficult. And that's not all, unfortunately multi axis JCMs can also be very hair pulling to set up and more than likely you'll need them for shoulder rotations. The scaling math can get pretty confusing and you'll need ValueOps to set it up. While the relations between a JCM that effects 2 positive rotations is tricky it becomes almost unbelievably confusing setting the scale values for a JCM that works against both a positive and negative rotation. I still haven't managed to get such a JCM to work properly, but I haven't bothered to post a query to the technical forum in hopes of getting Nerd to answer. He's probably the only one who can answer such a problem. My math mojo isn't up to the task it seems... even framing such a question coherently is a bit of work. Fortunately for the simple weightmapping projects I've been playing with I've been able to fix things with weight painting alone. If at all possible try to solve the most you can via the weights, it's much more artist friendly than diving into complex valueOps.

When you get to that point I really recommend finding a maths guru to help you setting these things up.

Best of luck Shane.