Governing the ever growing complexity of artificial systems on the one hand requires a number of expressive abstractions and different levels of interpretation, on the other hand suggests the adoption of formal / mathematical tools to (at least partially) model and predict the system behaviour. By adopting agent-oriented abstractions as the starting point, we argue that organisation, coordination and security all insist on the same conceptual space — that is, static / dynamic relations / interactions among agents —, which also represents one of the main sources of complexity for MAS, and for artificial systems in general, as well. The notion of ACC (agent coordination context) is used in this paper as the unifying core abstraction of a framework that encompasses all such issues, promoting the integration of organisation, coordination, and security. Such a framework, called RBAC-MAS, is expressed through a process algebraic model which integrates the classic organisational issues of role-based models (like RBAC) and the more recent works on interaction and coordination in MAS.