Audit binaries or low-level code for memory corruption risk, trust-boundary mistakes, dangerous parsing logic, and exploit-relevant mitigations. Use when Codex reviews native code or binaries for security weakness patterns.
