Propagating AUML Protocols to Detailed Design Yoosef Abushark ⋆ and John Thangarajah School of Computer Science and IT, RMIT University, Melbourne, Australia {yoosef.abushark,john.thangarajah}@rmit.edu.au Abstract. The interaction between agents is a key aspect of multi-agent systems. AUML sequence diagrams are commonly used to specify these interactions between agents in terms of interaction protocols. Whilst most of the popular agent oriented software engineering methodologies such as Prometheus, Tropos, O-MaSE, INGENIAS and GAIA support AUML protocol specifications in the design, the supportive tools do not provide any mechanisms for ensuring that the detailed design, and consequently the implementations, faithfully follow these protocols. In this paper, we show how AUML protocol specifications in the Prometheus methodology can be automatically propagated to the detailed design of the method- ology by creating appropriate artefacts. The approach is general to all design methodologies that follow the BDI model of agents. We empiri- cally show that the manual translation of protocols to the detailed design even for a simple AUML protocol can be a tedious and error-prone task for even relatively experienced users. The evaluation shows that our au- tomated approach address these issues to a large extent. Keywords: AOSE Methodology, Multi-agent system, Inter-Agent Interaction Protocols. 1 Introduction Intelligent Agent Systems are gaining popularity for building complex applica- tions such as Unmanned Aerial Vehicles [20] and Electronic trading agents [22]. Features such as autonomy, proactivity, flexibility, robustness and social abil- ity, are what makes these multi-agent systems (MAS) suitable for developing applications that operate in highly dynamic environments. However, these very features also makes developing and testing multi-agent systems a difficult and challenging task. A number of architectures have been proposed to developing MAS, in par- ticular, the popular Belief-Desire-Intention (BDI) agent architecture [21] where agents are developed using mental attitudes of beliefs, goals, plans, events, and so on. A number of agent oriented software engineering (AOSE) methodolo- gies have been proposed for designing and implementing systems based on the ⋆ Aknowledges King Abdulaziz University for scholarship. M. Cossentino et al. (Eds.): EMAS 2013, LNAI 8245, pp. 19–37, 2013. c  Springer-Verlag Berlin Heidelberg 2013