I think from a "fun game design" perspective, it's probably a terrible idea, but it still sounds like a fun exercise in of itself, because it will force you to design the mechanics in a way that's decoupled from the GUI, and thus easy to tweak.
I say it's probably a terrible "game design" idea because you run a very high risk of working hard on the mechanics, then working hard on the GUI, only to plug them into each other after much blood sweat & tears to discover that you've created a game that isn't fun. It's less of a risk since you're programming an RPG--a known concept that you can trust to have a certain appeal just by following a few conventions of RPG-ness--but in general, the best way to do game design is to "find the fun" first.
Like you say, RPGs are very much a known concept. That simplifies things a lot! I'm quite familiar with the genre, having spent plenty of time doing what I'll claim to be research on what makes a good RPG.
I'm modeling this game with various classic 8 bit RPGs in mind, so I have some idea of how the graphics work, as well as knowing what I personally find appealing about the game. Just think, if your favorite 8 or 16 bit RPG that never had a sequel got a sequel, what would it be like?
I say it's probably a terrible "game design" idea because you run a very high risk of working hard on the mechanics, then working hard on the GUI, only to plug them into each other after much blood sweat & tears to discover that you've created a game that isn't fun. It's less of a risk since you're programming an RPG--a known concept that you can trust to have a certain appeal just by following a few conventions of RPG-ness--but in general, the best way to do game design is to "find the fun" first.