Today's computer supported modelling environments could provide much more information about the users’ actions and problem solving processes than they usually store for later usage. Thus, relevant information about learning processes which could be used for reflection and analysis is lost. This paper describes an approach to tackle this issue by generating “reports”, in the sense of augmented summaries of states and action traces from modelling processes. This approach includes a) gathering information about actions and states from specific modelling environments, b) analysing these information using domain knowledge (if available) and c) represent the results in a way suited for various use cases like authoring, presentations or monitoring of learning processes.