We provide an update on dTank (Morgan et al., BRIMS 2005), a highly usable adversarial environment. It can be used for examining performance variability in situation awareness and architectural comparisons of competitive agents. First, the new design and implementation details of the updated dTank environment are discussed. In-progress models constructed with several cognitive and agent architectures (Java, Jess, and Soar) are then noted. Next, in the moderated behavior section, we present preliminary analyses of embedded performance delays in reaction to battlefield environmental conditions. Noise factors and variability in delay length at the tank commander level lead to different battle outcomes. Finally, we note some changes that will be required for dTank to better model situation awareness. Light-weight agent-construction environments such as dTank fill an important need for experimentation and prototyping tools that support quick scenario development and behavior implementations in a usable programming environment available to a wider user audience. These types of modeling tools can both raise and answer critical questions concerning agents' awareness of their surroundings and resulting behavior.