Insecure AI Agent Permissions
Critical · AI Security · OWASP Agentic
Description
Agents granted excessive API/shell access.
Root cause
Auto-approve all tool calls.
Impact
Mass data deletion, fraud, lateral movement.
Detection
Audit agent action logs.
Mitigation
Per-tool scopes; step-up auth; deny by default.
Secure coding
Treat agents as privileged users.
Safe lab example
lab-agent-perms
Practise related skills in Labs or deepen on Academy library.