Multi-agent systems, where independent agents collaborate to achieve complex goals, exhibit enhanced capabilities compared to individual agents working alone. Evaluating agent contributions is critical for improving system efficiency, debugging collaboration issues, and optimizing computational resources. Understanding agent roles, types, and collaboration patterns is essential for this evaluation. Techniques such as counterfactual testing, controlled variable testing, and statistical analysis can be used to measure agent impact in multi-agent workflows. Action Advancement metrics quantify an agent's effectiveness in making progress toward user goals, while hallucination detection metrics analyze factual consistency of agent outputs against known information sources. Galileo's implementation provides a robust contribution monitor that tracks performance metrics over configurable time windows, and real-time feedback systems for immediate interventions to optimize system performance.