You Aint Gonna Need It
Building truly generalized frameworks is hard. Put too little in, and there isn't a real advantage to the framework. Put too much in, and the investment to reach the really cool, really useful features (in terms of knowledge and experience) is too high and no one gets there. Finding the balance is a challenge.
HiveMind's core has been created to be as minimal as possible. Even generally (but not universally) useful things like the !EJBProxyBuilder service factory have gone into the standard library. It's very important to be certain that each change is worthwhile, and won't paint us into a corner. Occasionally, a use case pops up that is so universal, that it is worth changing the framework to match (for example, the <conversion> element that supplements and replaces the more powerful, more complex, more verbose <rules> element).