“Security scanning”

AGENTS.md from openai/codex-security: 149 lines covering Deep Scan worker settings, Avoid speculative defenses, Preserve diagnostic text, Public CLI changes…