VUL-Rag: LLM-based Vulnerability Detection

Discipline: CS

Advisor: Dr. David Mohaisen

This project is a novel vulnerability scanning engine designed to detect vulnerabilities that often evade traditional rule-based and heuristic detection. By leveraging a large knowledge database and AI, the system aims to improve the accuracy of vulnerability identification. We provide a web application where developers can upload their code for scanning and receive the engine's vulnerability analysis. Our database currently contains over 8,000 different vulnerabilities. This project has the potential to shift traditional vulnerability scanning away from a purely semantics-focused approach where only the exact text of a code base is examined toward an understanding-focused approach, where the engine analyzes what the code is doing and how that behavior could be linked to a vulnerability.

Back to Showcase