The unreasonable effectiveness of LLMs for auditing Rust code | Dark Hacker News