A novel method for predicting software vulnerabilities is presented in this paper. It examines the classes, methods, and functions that comprise software using static analysis. By incorporating this analyzed data into machine learning models, the technique can identify any errors. When tested on publicly available datasets, our strategy performed better than earlier approaches. Engineers may concentrate on fixing significant defects as a result of this increased efficiency, which ultimately strengthens software security. By employing this technique, software development teams can proactively build more secure systems and reduce the likelihood of exploitation.