Fault localization and service restoration (FLSR) is a crucial aspect of managing power distribution systems’ fault, aiming to enhance resilience and service reliability. Various methods have been developed to address the FLSR problem, including centralized and distributed approaches. In Tanzania, fault management in the secondary distribution network (SDN) is handled manually, primarily through customer reports and on-site inspections. The decisions regarding the service restoration depend on past experiences, the rated transformers’ capacity, and peak hour demand, which often leads to extended service restoration time and load shedding. This study used the Python Agent DEvelopment (PADE) framework to develop a multi-agent system (MAS) for FLSR, considering operational constraints and load shedding. Actual interaction of the software agents implemented in Raspberry Pis with a real power system through an intelligent electronic device has been performed. Four agents have been developed to assist the FLSR process: grid agent, control agent, load agent, and switch agent. The multi-agent-based approach facilitates decision-making by localizing faults and enabling restoration through load transfer to the nearby grid agent and load shedding. The computation time for deployed MAS in Raspberry Pi was 4.124 s. The future work will focus on the actual deployment and testing of the deployed algorithm on a real large power system and the implementation of deep learning algorithms for load shedding.