Aim to mention 20-40 concrete checks. This expanded list helps you turn a short answer into a structured senior-style response.
Each role can access allowed features and cannot access forbidden features.
Admin-only actions, owner-only actions, member actions, guest actions, and read-only restrictions work.
Hidden UI actions, disabled controls, and access denied pages match permission rules.
Role upgrade, downgrade, removal, conflicting roles, and multiple roles update access correctly.
Existing sessions, cached permissions, expired tokens, and revoked tokens reflect permission changes.
Last admin protection and role assignment rules prevent unsafe account states.
Direct URL access and direct API access are denied when unauthorized.
Tenant isolation, workspace isolation, IDOR, payload tampering, mass assignment, and privilege escalation are blocked.
401, 403, safe error responses, CSRF, XSS, SQL injection, open redirect, and rate limiting are verified.
Search, filters, pagination, exports, and reports do not leak restricted data.
Role changes create audit logs, metrics, alerts, and useful diagnostic logs.
Concurrent role changes, transaction consistency, permission cache invalidation, and performance are tested.
Structure your answer: Start with a role matrix: each role, allowed actions, forbidden actions, and expected UI state.
Structure your answer: Add backend authorization: direct URL, direct API, status codes, and safe errors.
Structure your answer: Add permission changes: upgrade, downgrade, revoked token, cached permissions, existing sessions.
Structure your answer: Add security and data leakage: tenant isolation, IDOR, mass assignment, privilege escalation, audit logs, and concurrency.
Do not miss: Direct API access when UI controls are hidden.
Do not miss: Tenant isolation, workspace isolation, and IDOR.
Do not miss: Cached permissions after role changes.
Do not miss: Audit logs and last admin protection.