Jev and CodeGraph for security review
The typesafe-security-review project combines Jev’s probability judgments with a code graph and OWASP/CWE data for low-cost security scanning.
The post notes that TypeSafe's own docs cover Jev prompt injection, advising allowlisting, sending only required fields, and pinning injected fixtures so a flipped verdict won't break upgrades.
Jev prompt injection is in TypeSafe’s own docs. Allowlist first, send only needed fields, pin injected fixtures so a flipped verdict blocks upgrades. https://automater.ai/intel/jev-gate-injection-jaggedness/…