Department of Computer Science, K.U.Leuven, Leuven, Belgium
CW Reports vol:CW321
Abstract Programming agents is a hard task, mainly because of the diverse aspects of an agent that need to be considered. A developer has to decide how the functionality of the agent is implemented, in what form the agent will store its information and how it will handle the sometimes complex interactions with other agents. Additionally, there are the many security requirements to take into account. Access to agent services needs to be controlled, network communication should be protected, accounting must be performed, agent data need to be protected, etc. The current agent platforms do not really support these aspects very well. As a result, the agent programs become very tangled and thus are hard to maintain, evolve and reuse. In this paper, we present an open agent platform, targeted at experimentation, both in application structure and security. The platform starts from a more explicit model of an agent and its platform. We argue that this model allows more reuse and flexibility in the construction of agent applications.