As the use of hand gestures becomes increasingly prevalent in virtual reality (VR) applications, prototyping Compound Freehand Interactions (CFIs) effectively and efficiently has become a critical need in the design process. Compound Freehand Interaction (CFI) is a sequence of freehand interactions where each sub-interaction in the sequence conditions the next. Despite the need for interactive prototypes of CFI in the early design stage, creating them is effortful and remains a challenge for designers since it requires a highly technical workflow that involves programming the recognizers, system responses and conditionals for each sub-interaction. To bridge this gap, we present GestureCanvas, a freehand interaction-based immersive prototyping system that enables a rapid, end-to-end, and code-free workflow for designing, testing, refining, and subsequently deploying CFI by leveraging the three pillars of interaction models: event-driven state machine, trigger-action authoring, and programming by demonstration. The design of GestureCanvas includes three novel design elements — (i) appropriating the multimodal recording of freehand interaction into a CFI authoring workspace called Design Canvas, (ii) semi-automatic identification of the input trigger logic from demonstration to reduce the manual effort of setting up triggers for each sub-interaction, (iii) on the fly testing for independently validating the input conditionals in-situ. We validate the workflow enabled by GestureCanvas through an interview study with professional designers and evaluate its usability through a user study with non-experts. Our work lays the foundation for advancing research on immersive prototyping systems allowing even highly complex gestures to be easily prototyped and tested within VR environments.