This document discusses using a multi-agent system to provide decision support for ethical problem solving. It reviews normative and descriptive theories of ethics and proposes four roles for decision support agents: advisor, group facilitator, interaction coach, and forecaster. The roles involve comparing user decisions to normative ethics, facilitating group processes, helping individuals understand how others may react to their ideas, and predicting potential outcomes of solutions. The goal is for computer-based agents to provide ethics support to accompany users wherever and whenever tasks are performed.