The demand for assessments of competencies that require complex human interaction is steadily growing as we move toward a focus on twenty-first century skills. As assessment designers aim to address this demand, we argue for the importance of a common language to understand and attend to the key challenges implicated in designing task situations to assess such competencies. We offer the descriptors "coordinated," "accumulated," and "dynamic" (CAD) as a way of understanding the nature of these competencies and the considerations involved in measuring them. We use an example performance task designed to measure teacher competency in leading an argumentation-focused discussion in elementary science to illustrate what we mean by the coordinated, accumulated, and dynamic nature of this construct and the challenges assessment designers face when developing performance tasks to measure this construct. Our work is unique in that we designed these performance tasks to be deployed within a digital simulated classroom environment that includes simulated students controlled by a human agent, known as the simulation specialist. We illustrate what we mean by these three descriptors and discuss how we addressed various considerations in our task design to assess elementary science teachers' ability to facilitate argumentation-focused discussions.