3 ms·Benchmarking Vulnerability of Agent-Generated Code in Real-World Tasks1 points by doppp 10mo ago