Sub-Patterns

01

Pre-Execution Disclosure

Before the agent starts

Disclose what the agent will do, what data it will access, and under whose authority it is operating. If the agent requires access to systems or data not previously authorized, a consent gate is required before it can proceed.

Required components

02

Active Execution State

While the agent is running

Surface continuous state: what the agent is doing right now, how far along it is, what action comes next. The monitoring layer watches for anomalies — behavioral drift, scope escalation, unresolvable decision forks — and surfaces attention triggers independent of what the agent itself reports.

Required components

03

Approval Gates

Before irreversible actions

Any action that is irreversible, high-consequence, or outside the agent's original authorized scope requires an explicit human approval gate before execution. Risk level determines whether the gate is mandatory or advisory.

Required components

04

Post-Run Audit

After execution completes

Every agent run produces an immutable, exportable audit trail — timestamped, typed, and tied to human approval events. The impact assessment surfaces the full footprint: what data was touched, what systems were affected, what regulatory exposure was created.

Required components