The management of software projects is a critical activity and susceptible to unplanned situations, commonly known as risks. Risks stem from a variety of sources, both external and internal to the project or organization; moreover, they can occur at any stage of the project life cycle. In this paper, we will present an approach that provides support for identification, analysis, response planning and risk control in software projects. For this purpose, an agent was developed and its behavior is based on metrics, change requests in the project, as well as the use of contingency reserves. The risk agent, ARis, was inserted into an existing multi-agent system (MAS) and in conjunction with all the agents, assists the prediction and mitigation of risks in software projects.