This paper presents the design, development, and implementation of a Multi-agent System (MAS) aimed at solving a simulated treasure hunt problem in a distributed environment. The system incorporates two distinct types of agents—explorer and collector agents—that work collaboratively and in coordination to navigate an unknown map, locate hidden treasures, and efficiently gather them. Explorer agents are tasked with dynamically mapping the environment and unlocking treasure locations, while collector agents use this information to plan optimal routes for treasure retrieval based on specific constraints such as treasure type and backpack capacity. The MAS is developed using the JADE framework for agent communication and interaction, while the Dedale environment provides the infrastructure for testing agent behavior in complex, multi-node map scenarios. A key aspect of the system is the use of indirect communication via a broker agent, which manages the exchange of information between agents to optimize resource use and avoid direct communication overhead. The architecture, communication protocols (FIPA-ACL), and coordination mechanisms are described in detail, highlighting how rationality, reactivity, and cooperation properties are incorporated into agent behaviors. The performance of the system is evaluated through various metrics, including task completion time, communication efficiency, and the number of movements required to complete the treasure hunt. This study contributes to the field of multi-agent systems by demonstrating how distributed agents can solve complex problems through collaborative and coordinated actions in deterministic environments, with implications for applications in logistics, resource allocation, and autonomous systems.