OpenAI just released “codex security”, worth trying (along with other suggestions) if your org has access https://openai.com/index/codex-security-now-in-research-prev...