Automated Security Analysis of Dynamic Web Applications through Symbolic Code Execution

摘要

The automatic identification of security vulnerabilities is a critical issue in the development of web-based applications. We present a methodology and tool for vulnerability identification based on symbolic code execution exploiting Static Taint Analysis to improve the efficiency of the analysis. The tool targets PHP web applications, and demonstrates the effectiveness of our approach in identifying cross-site scripting and SQL injection vulnerabilities on both NIST synthetic benchmarks and real world applications. It proves to be faster and more effective than its main competitors, both open source and commercial.