Abstract
The aim of this research is to equip agents in a Multi-Agent System (MAS) with reusable autonomous capabilities to enable them to behave in unknown situations and complex simulated environments. A concept demonstrator called Web-based Multi-Agent System Testbed (Web-MAT) is developed with a context switchable capability currently implemented as a Real-Time Strategy (RTS) game. The demonstrator is intended to enable researchers to improve/develop the communication aspects within an agent-oriented architecture and to program agents that automatically adapt functionality at runtime using event stimuli. Complementary research on a variety of agent tools (specifically JACK, JADE and CIAgent) and communication languages (ACL, KQML, FIPA and SOAP) has been reviewed to glean knowledge that enables these agents to adapt into Web-MAT and generate encouraging results.