NovGuard Research
Practical AI security references.
Turn risk into concrete tests, controls and evidence.
AI Agent Security Checklist
Permissions, tools, secrets, approvals, memory, retrieval and incident evidence.
Open →Prompt Injection Defense Matrix
Map injection paths to provenance, authorization, output and runtime controls.
Open →MCP Security Controls
Server trust, permissions, arguments, data movement, approvals and auditability.
Open →