I think it will turn out that the definition of "good" is a combination of recursion ("things which encourage the growth of good things") and opposition ("anything that helps prevent bad things from happening") -- where "bad" is actually the easier term to define.
Not that it's easy, just easier... my first guess would be something like "that which increases the entropy/functioning/? of society" (killing or harming people, disrupting the supply of basic needs...) but I wouldn't try to claim that it's that simple.
it's important to note, however, that very few actions are unambiguously one or the other, because something which has good consequences in the short term may have long term bad consequences, or vice-versa.
I propose that any rational moral argument is ultimately about that question (rather than about what "basic good" and "basic bad" are): does action X, which seems clearly good now, have clearly bad consequences later that outweigh that good?
We also get into trouble when we lump actions together with their canonical (rather than reasonably expected or _actual_)* consequences -- the goodness-or-badness of the expected consequences becomes the goodness-or-badness of the action, even if the action was done in such a way as to create a different outcome that is unexpectedly better/worse than the model assumes.
Okay, I've rambled enough... bracing for impact...
(* edited for clarity)