This is an odd one.
In general, I'd echo freeformschooler's advice that when in doubt, just go for it. Practice makes perfect, and all. Plus, sometimes the only way to figure out what you're after is to actually see it in action.
More specifically, I tend to go for more mechanical, less RP-y games, so my expertise is of somewhat limited use here, but I can offer a few suggestions.
First of all, if possible, I'd suggest getting a default resolution mechanic that covers every possibility. This can be as simple as "the GM makes something up" or "attribute check using whatever's closest to relevant," and it can be overridden whenever necessary, but having something in place for those things you never thought of can be handy. Trust me, players will get weird.
Secondly, try to have concrete mechanics in place for things that really matter. Lacking clear, hard numbers is an invitation for something to either not matter at all, or be too vague for players to really understand the consequences of.
Third, just spend some time considering what you want to happen or how you want certain things to work, and figure out how the mechanics or GMing style should reflect that. Actions that give loot or experience tend to be more intrinsically rewarding than those that don't, for instance, so you'll tend to see behaviors leaning in those directions where there's a difference.
Hope this helps.