The concepts and methodology needed for designing, developing, and implementing real life applications based on multi-agent systems are today still a challenge for researchers in Artificial Intelligence and Computer Science. Industrial-strength multi-agent systems require, among other things, reusability, i.e. the capability of not having to design and implement from scratch a new multi-agent system for every new application domain. In order to improve the reusability of multi-agent systems, we have taken a knowledge modelling stance coming from the research on knowledge engineering, and adapted its insights to deal with multi-agent systems. The resulting contributions are organized in the Open, Reusable and Configurable Agent Systems (ORCAS) framework, comprising conceptual guidelines to design, methodological commitments for development, and an implemented platform to make concrete these abstract notions and effectively experiment with them. This paper describes the ORCAS e-Institution, and agent platform for developing and deploying open, reusable and configurable cooperative multiagent systems.