OpenAI's security agent finds, checks, and fixes code weaknesses.
It is a building inspector with a screwdriver. It jiggles the lock, spots the wobble, and calls the handyman.
You meet it in GitHub security reviews. It checks bugs and suggests patches. A human still reviews before merge.
Codex
It extends Codex's coding skills into app security.
AI Vulnerability Discovery
It finds complex weaknesses and suggests fixes.
AI Code Review
It moves code review from style checks to security checks.
AI sandbox
The sandbox lets it test bugs without touching the real system.