A taint-based call graph construction algorithm for Java object deserialization.
A dataset of prompts to evaluate ML-based code generation models with respect to their ability to generate secure code.
A dataset of prompts to evaluate ML-based code generation models with respect to their ability to generate regular expressions.
A dataset of prompts and automated framework to evaluate ML-based code generation models with respect to their ability to generate functional and secure code.
We extend the original SALLM dataset of English-only prompts into a multilingual dataset covering 23 natural languages and two more programming languages (Java & C++).
A multi-agent framework for improving the precision of Java static vulnerability detection.