Why? Most teams ship AI features without security testing. This action catches prompt injection, jailbreaks, and data leakage before they hit production.
How it works: - Add 5 lines of YAML to your workflow - Scans run automatically on push/PR - PRs get blocked if critical vulns are found - Full report with remediation steps
Free tier: 5 scans/month. 650+ attack vectors.
Built this because I do AI red teaming professionally (OSCP+, C-AI/MLPen). Happy to answer questions!