@drwhax I think LLMs get the low hanging fruit of vulnerabilities, so you shouldn't trust an LLM's security audit as a definitive audit. As models get better, they'll catch higher up fruit, but still.