Introducing CaT, a minimalist algorithm for reinforcement learning, effectively enforcing constraints in legged locomotion tasks.