The traditional agent architecture is essentially based on three major components, namely, its competences, its roles, and its interactions. In a previous work, we proposed to dissociate the interactions from the agent content, and then to consider them as separated and shared resources among agents. In this way, multi-agent systems are called Interaction-Centered MAS where interactions are easily updated, maintained and reused independently without changing the agent source code. In this context, the system requires an Automatic Reuse of Interaction Protocols (ARIP). In this paper, we describe the ARIP model, its new concepts (like protocol description schema, agent description schema, etc.), and its main functionalities making possible the reasoning on the interaction protocols and their automatic reuse.