The growing reliance on Internet of Things (IoT) systems for smart homes, cities, industries, and healthcare has made resilience a critical concern. MQTT, a lightweight messaging protocol, is widely adopted for these systems, but failures-particularly at the publisher level-can lead to cascading system disruptions. This paper presents the design and implementation of a chaos injector framework that mimics the concept of vaccination in biological systems by introducing controlled chaos into MQTT-based IoT systems. Our approach autonomously scans MQTT topics and performs replay attacks, simulating data failures at the publisher level to strengthen the overall resilience of the system. The chaos injector acts as a "vaccine" for IoT, enabling the system to recognize, respond to, and recover from real-world failures. By testing the tool in a real smart office deployment, we uncovered system weaknesses and identified the underlying causes. Through this work, we contribute to the development of more fault-tolerant and resilient MQTT infrastructures using chaos engineering..