Prototype B · M4

Penjelajah Keamanan & Kebijakan

Simulasi fail-closed untuk Identity → Role → Permission → Authority → Consent → Context → Policy Decision. Semua data di halaman ini bersifat blueprint/demo dan tidak terhubung ke production identity provider.

Policy input

IdentityAction Mandate valid Authority allows action Consent valid

Policy decision

Decision

ALLOW

POLICY_SATISFIED

Permissions: case.read, payment.read, reconciliation.read, reconciliation.suggest

Data classes: INTERNAL, CONFIDENTIAL

Pagar pengaman keamanan

Every human, agent and service has a distinct identity.
Agent identities never inherit human roles or approval privileges.
Permission is necessary but not sufficient: mandate, authority, consent, context and segregation-of-duties are also evaluated.
Human Only permissions are denied to all agent identities.
Maker and checker must be different human identities for the same approval request.
Access to CONFIDENTIAL or RESTRICTED data requires explicit classification allowance.
Automation orchestrators can request allowlisted actions only and cannot own or directly mutate business state.
All policy decisions are auditable and fail closed when context is incomplete.