The capability of agent-based modeling in the Mesa framework to provide a novel method for detecting malicious URLs is highlighted in the proposed model. As phishing assaults continue to pose a cybersecurity threat, efficient URL classification techniques are essential. The Multi-Agent Systems (MAS) technique, implemented with Mesa, achieves strong performance in differentiating between phishing and authentic URLs by training a Random Forest classifier on features. Comprehensive testing demonstrates the framework’s efficacy and highlights the contribution of the Mesa environment to improving URL threat detection capabilities. It emphasizes how important Mesa’s agent-based modeling is as a potential means of enhancing cybersecurity protections against fraudulent URLs.