Given that SQL injection attacks continue to pose a substantial threat to the security of web applications, this paper critically assesses sophisticated security measures intended to mitigate these vulnerabilities. We investigate the Agent-based Vulnerability Response System (AVRS), which improves traditional intrusion detection systems by incorporating mobile agents that provide increased autonomy and mobility. This system integrates a comprehensive vulnerability database and machine learning techniques to enable real-time threat detection and response. The VIWeb vulnerability scanner is introduced in the study, which evaluates three machine learning models—Decision Trees, Support Vector Machines (SVMs), and Artificial Neural Networks (ANNs)—for malware detection. The scanner employs algorithms such as the Reverse Resemblance Algorithm and Malicious String-Matching Algorithm. According to performance metrics, ANN surpasses SVM and Decision Tree approaches in its ability to classify threats, achieving the highest accuracy (86.50%) accurately. The results emphasize the potential of integrating machine learning with conventional security measures to fortify defenses against SQL injection assaults, thereby establishing the groundwork for future research and implementation strategies.