This is so wrong, I don't even know where to begin. The _entire_ point of URLs is that they're just that -- Universal. Requiring a specific application in order to consume a URL is just as asinine as User Agent sniffing to enforce an 'IE Only' policy (and you're lying if you say that ended well).
Perhaps consuming based on Content-Type headers would be a better way to attack this (legitimate) problem. Certainly in the early days of the web, that's how it was done (want to enjoy the latest VRML? Install plugin 'x'). Adding a layer of abstraction on top of URLs that tries to mediate this is just altogether wrong, I think.
Debate welcome. I'm mostly just thinking out loud here.