THE HANDPROTOCOL
System nominalLive demo
SECURITY & GOVERNANCE

Every action meets the gate.

Deterministic policy remains between intelligence and execution.

01

RISK MODEL

Five levels.
One hard stop.

R0

Observe

Read-only insight

R1

Maintain

Reversible action

R2

Communicate

Bounded external write

R3

Human required

Sensitive authority

R4

Forbidden

No approval can unlock it

02 / DECISION BOUNDARY

Capability stops where policy says no.

AEGIS does not interpret power as permission. Every request resolves to one explicit outcome before execution.

ALLOWExact, current and low-risk policy
HUMANSensitive authority requires review
BLOCKForbidden or undefined action
A central machine between luminous and shadowed councils as a metaphor for policy evaluation
03 / FORBIDDEN

Some authority should never exist.

Change ownership
Grant autonomous privilege
Delete audit history
Impersonate GHOST
Take on debt
Disable human oversight

See AEGIS make the call.

Allowed, human-required and categorically blocked decisions are all present in the deterministic demo.

Open permission engine