We argue that for the purpose of evaluating and categorizing dexterity, a richer taxonomy than the canonical grasp postures is needed. We will augment the taxonomy of grasp postures with affordances of what is actually done with those postures, by performing a systematic identification via surveys of the literature, our testbed teams, and our ecosystem partners. We further will identify metrics and benchmark tasks associated with each grasp-affordance combination, and create benchmark simulation environments.