This is a review paper on multi-agent systems with focus that represents users in various tasks performed in a heterogeneous environment. This is a contentious topic that attracts many researchers and repels others. It represents a research frontier in a number of fields including computer science, artificial intelligence, robotics etc. Multi-agent system has been an extreme active research area since 1992, and now has raised considerable interest in the research community and in industry. But the long promised deployment has not materialized yet. This review paper looks back at the history, present status and potential future of multi-agent systems on critical view base. It consists of four chapters: Chapter 1 focuses on multi-agent system modeling; Chapter 2 on multi-agent systems architecture, Chapter 3 on multi-agent system applications, platforms and development tools, Chapter 4 gives a brief summary over multi-agent system history during the last 10 years, the present development status and future direction. This report has demonstrated that multi-agent system is still far away from complete. It expects a convincing story from either the artificial intelligence society or software engineering community. However, this does not prevent people from believing that agents systems would play significant roles in many application domains including software engineering, computer network, robotics, automated manufacturing, etc.