Abstract
Multi-Agent Systems provide an ideal level of abstraction for modelling complex applications where distributed and heterogeneous entities need to cooperate to achieve a common goal, or to concur for the control of shared resources. This paper proposes a declarative framework for developing multi-agent systems. A formal approach based on Logic Programming is proposed for the specification, implementation and testing of software prototypes. Specification of the PRS agent architecture is given as an example of application of our framework.