Just read Models and Modelling in Economics by Mary S. Morgan and Tarja Knuuttila in U. Mäki (ed) Handbook of the Philosophy of Economics. While nearly all the examples are taken from economics and while it considers all sorts of models, not just agent-based models, I found it very interesting and thought provoking.
Excerpts from the conclusion section say it well:
… philosophers of economics have aimed to analyse how economic scientists build models and what they do with models: how they use them, how they argue with them, and what they learn from using them.
This essay has examined a set of the issues that have emerged from this work and that in many respects beg for further analysis: the problems of de-idealization and what these say about idealization; the implications of models conceived of as fictions, artefacts and mediators; the different ways in which models are taken to represent and mimic; the importance of how models are used and thus their experimentable potential; the roles of content and materials in providing resources and constraints to modellers; the functions of stories, analogies, templates, credible world comparisons, and statistical rules in making and supporting different kinds and modes of inferences; and so forth. These various new foci are both distinctive in terms of topics, and thought provoking, if not challenging, to the older conventional philosophical positions. They follow, however, not just from a naturalistic turn towards the study of science, but also from a reframing of the basic object of study: from models to modelling, that is, to how economists construct models and work with them.
The message I take home from this for the Modelling4All project is that there is a huge variety of different conceptions of what models are, of why one might construct them, and how one might use them.