Atlas uses its whole body — legs, arms, torso — to perform a sequence of dynamic manoeuvres that form a gymnastic routine. We created the manoeuvres using new techniques that streamline the development process. First, an optimisation algorithm transforms high-level descriptions of each manoeuvre into dynamically-feasible reference motions. Then Atlas tracks the motions using a model predictive controller that smoothly blends from one manoeuvre to the next. Using this approach, we developed the routine significantly faster than previous Atlas routines, with a performance success rate of about 80%.