Cooperative software agents for the Unix domain

摘要

The paper presents an agent architecture that integrates multiple capabilities into a single intelligent agent allowing it to interact both with other agents through an expressive agent communications language and with humans through a human agent interface, and to satisfy external requests either through the direct execution of a set of actions or through a plan defined by an internal planner. This agent architecture has been implemented using a programming environment which takes advantage of object-oriented programming features, and has been experimented with in the Unix domain specializing it to define an agent able to perform a set of Unix commands and to execute actions in cooperation with other agents.