Service-oriented architecture (SOA) provides a flexible paradigm to compose dynamic service processes using individual services. However, service processes can be vastly complex, involving many service partners, thereby giving rise to difficulties in terms of pinpointing the service(s) responsible for problematic outcomes. In this research, we study efficient and effective mechanisms to deploy agents to monitor and detect undesirable services in a service process. We model the agent deployment problem as the classic Weighted Set Covering (WSC) problem and present agent selection solutions at different stages of service process deployment. We propose the MASS (Merit based Agent and Service Selection) algorithm that considers agent cost during QoS-based service composition by using a Meritagent cost heuristic metric. We also propose the IGA (Incremental Greedy Algorithm) to achieve fast agent selection when a service process is reconfigured after service failures. The performance study shows that our proposed algorithms are effective on saving agent cost and efficient on execution time.