PERMISSIONS AND ACCESS CONTROL

PRACTICE LAB CONNECTION In the Resume Review Agent lab (Chapters 65 to 67), you will add a least-privilege permission gate: a resolver with default deny, an allow-list of tools and folders, and an approval step for sensitive actions. Test the prompts below against the permission configuration you build. ARTIFICIAL INTELLIGENCE ASSISTANT PROMPTS [START OF PROMPT] Review this agent's current permission configuration and reduce it to least privilege: for each granted permission, state whether it is actually required for the agent's job and, if not, the exact restriction to apply: [PASTE LAB OR PRODUCTION PERMISSION CONFIG / ALLOW-LIST HERE]. Output format: a table (Permission, Needed? Y/N, Justification, Restricted Version if applicable). [END OF PROMPT] [START OF PROMPT] Review our secret-handling approach for these externally authenticated tools and flag any place a key or token could leak into the model's context or into logs: [PASTE LAB OR PRODUCTION SECRET-INJECTION CODE / CONFIG HERE]. For each leak risk found, give the exact fix, and specify what should happen automatically if a key is ever exposed. Output format: a findings table (Location, Leak Risk, Fix) plus an incident-response note for key exposure. [END OF PROMPT]