The Internet of Things (IoT) enables a network of resource-constrained devices to interact with the physical environment with minimal human intervention. In smart homes, managing gas leakage presents significant challenges due to the system’s heterogeneity and complexity, including issues related to contextualization, coordination, autonomy, and advanced reasoning. This work proposes a multi-agent gas detection framework that integrates IoT, multi-agent systems, and knowledge graphs to address these challenges. To validate the proposed framework, a set of execution scenarios are presented to show how the system responds to different gas leak situations.