security-hardening

AIDefence security layer with prompt injection blocking, input validation, sandboxed execution, output sanitization, and STRIDE threat modeling.

Independent GitHub snapshot · Automated classification · Original documentation governs