The usual way to design a simulation of a given phenomenon is to first build a model and then to implement it. The study of the simulation and its outcomes tells if the model is adequate and can explain the phenomenon. In this paper, we reverse this process by building a browser in simulations space: we study an automatically built simulation to understand its underlying model and explain the phenomenon we obtain. This paper deals with automated construction of models and their implementations from an ontology, consisting of generic interactions that can be assigned to families of agents. Thanks to the measurement tools that we define, we can automatically qualify characteristics of our simulations and their underlying models. Finally, we offer tools for processing and simplifying found or existing models: these allow an iterative construction of new models by involving the user in their assessment. This simulation space browser is called LEIA for "LEIA lets you Explore In- teractions for your Agents".