Live coding has typically focused on textual interfaces to interact with code, but this reliance on typing text impairs the usability of live coding as a performance practice. VJing practice is limited by a lack of fine-grained content control and a number of audiovisual tools are limited by their ability to be reused or shared in new performances. We propose the code jo࠻ey practice (CJing), a new hybrid practice where a performer known as a code jo࠻ey (CJ) improvises, manipulates and composites code to generate procedural graphics. CJing combines elements from live coding and VJing by offering user interfaces that abstract upon the code and provide high level interactions. To illustrate CJing we introduce Visor, a live coding environment for producing graphics in audiovisual performances. We present a typical Visor performance scenario and discuss Visor’s underlying implementation details.