Sure, just do:
Sprite.frame = 2;
Where 2 is whatever frame you want to show. This works for sprite sheets. If you're using a Texture Atlas and the frames have names, you'd use Sprite.frameName instead.
You can ask, plicatibu, but I might not be able to answer questions or help much for awhile. It was just last week we got the patches in to get Phaser working with CocoonJS more or less cleanly. And I've been spending most of this week working on gamepad code for the Ouya myself.
It's on my TODO list to document the IAP and ad code in CocoonJS, though, but I might not get there for a couple of weeks or more.
So this week has been pretty monumental. I started it on anti-biotics and feeling like hell thanks to a chest infection, but progress has still been rapid to say the least.
All of the old Game Objects have now been moved over to 1.2 and given a serious spring clean. Gone are the strange inheritance structures and they're all now clean Object extensions, usually from either Phaser.Image or a core Pixi object. This means lots less duplicate code and in some cases loads of new features for less KB than before. For example the Phaser.Text class is much cleaner now while supporting some nice new things like mouse over events, drop shadows, gradients fills and more.
World and Stage have gone on a diet too and a huge amount of the physics is finished. Outstanding on this are easy ways to set-up collision groups (it's all bitmask driven, so I want to find a way to handle that nicely), some World properties and finally adding in Springs (tempted to maybe leave this for 1.2.1).
There's new BitmapFont and BitmapText classes too. BitmapText handles a lot of things more cleanly now (fonts with spaces in the name for example! woot!) and BitmapFont is for when you want to use old-school sprite sheet style fonts SpriteBatches are in now too.
The most complex problem left to solve is tilemap collision. I'm more than happy with how it's working internally and found a nice way to speed-up rendering considerably, but still need to find an elegant way to hook it up with p2. Most likely walking the collideable tiles and building a polygon from it, but this will need testing. The physics objects do have AABBs, so I could possibly use those for tilemap collision - but it won't be as accurate and I really want all the things you'd expect from it. More experiments needed.