Codex Security
About
Security review rarely keeps pace with code changes, and static scanners produce noise that nobody remediates. Codex Security applies an agent loop to that gap: discovery runs scan a codebase, each candidate finding is validated before reporting, generated patches are verified, and verified fixes become GitHub pull requests for human review. The CLI wraps a TypeScript SDK so teams can embed scans in CI with an API key or authenticate interactively through ChatGPT, with Trusted Access gating for sensitive finding categories. A findings service stores results in SQLite, deduplicates them by embedding similarity, and serves a dashboard. Security teams and maintainers use it to move from scanner output to verified remediation, and Docker Compose configurations support bulk scans across repository sets.
Other entries by openai
- Codex CLI — agent