Verve is a library implementing general purpose reinforcement learning agents. It combines radial basis functions, temporal difference learning, planning, uncertainty estimations, and curiosity. It is intended to be an out-of-the-box solution for roboticists and game developers.