I have a proof of concept of input scope but it does cause some minor issues. I'm not sure if it is worth it. A Textbox has opacity of zero but is placed on top of the password box. You bind the password's value to the input box.
What you gain is input scope, what you lose is the cursor and the password 1 char "look ahead" (aka *i -> **x style)
Your second question should really be a new thread. Back buttons will cancel out of the prompt. It will cause the prompt to fire off the completed event but you can determine it is a cancel event by the
via something like: e.PopUpResult ==
If you look at the prompts.xaml.cs in the test application, you can see examples of this.