Architectural considerations about open distributed agent support platforms

Agents are becoming one of the most important topics in massively distributed and autonomous decentralized systems. Here, agents are considered as software computational entities, created and supported through agent platforms and acting in an agent environment. Based on some recognized agent skills and roles, the paper presents a discussion about the capabilities required in order that a standardized platform supports the action of agents in distributed heterogeneous systems. After posing some architectural goals and principles, a proposal is presented for an agent platform architecture that offers a set of configurable functionalities aiming to support generic agent necessities within their execution environment. The proposal follows the concepts of agent facilities described for Corba environments