Simulations provide an environment to experiment safely, openly, and repeatedly for learning mastery. However, many simulation environments experienced within a classroom fail to include the assessment components meaningful for instructor interpretations in a way that translate to standardized "scores". Even when a measure of standard assessments is included, often it fails to account for the unpredictable nature of decision-making within the complex, 3D, open-ended simulation environment. Embedding assessments within a virtual simulation environment poses several issues. First, the program must provide assessments that will fulfill educational requirements that will not take the learner cognitively "away" from their activities. Second, the program must provide an engaging gamelike experience for educational purposes. Third, it must provide assessments that maximize the unique capability inherent within digital deliveries, that allows for geographically disparate and asynchronous schedules between instructor and learner. This study addresses each of the above concerns through an integration of the classroom requirements and simulation affordances. Through the inclusion of a "replay" function for self-regulated after-action review, students answer questions about their understandings, but also thoughtfully reflect on their process and the applications for those understandings in novel scenarios. Created within an educational curriculum, S'cape is designed to function as a stand-alone module to teach and evaluate understandings about core concepts. This platform was piloted as a first person explorer game addressing various levels of complexity about chemical and physical properties of substances. Refined within gaming and technology best practices, this novel architecture combines educational and gaming principles. Engaging learners through the use of the automated assessment features (i.e., automated embedded assessments as after action review (AAR) and a ‘replay' function for metacognitive support), in this case, holds promise for military and corporate scenarios to aid in the valid systemic needs of complex, open-ended assessments. The paper describes how automated embedded assessments may provide a means for safe experimentation while supporting metacognitive practices crucial for 3D training environments.