dacuma-labs/agent-security-audit
Skill Claude CodeCodex
Audits AI agent projects for security risks in prompts, tools, memory, RAG, and permissions. Use when: the user asks to review agent security, harden an agentic system, check for prompt injection risks, audit tool permissions, or secure an AI assistant before deployment. Do NOT use for: projects with no agentic…
2 5mo ago A 82 tokens
original MIT