Containment Assurance · Assessment Instrument v1
The Containment Audit
Sandbox & agent isolation — a verdict on whether the box actually holds
Answer each check about your own system — Yes, Partial, or No. The instrument scores six containment layers by severity and returns a verdict you can act on: which layers are sound, which are decorative, and exactly where the exposure is. The standard throughout is one principle — the contained thing must never enforce its own containment.
The Standard
The contained thing must never enforce its own containment. Every weak result below is a version of that single failure — filtering inside the process, cooperative limits, self-written logs, trust-based isolation. Sound containment is enforced from outside the contained thing, default-deny, and witnessed by something the contained thing cannot touch. A boundary enforced by the inside is not a boundary — it is a request.

Assessment Report

0%
INCOMPLETE
Answer the checks to generate a verdict. 0/24 answered.
CONTAINMENT AUDIT v1 · 6 LAYERS · 24 SEVERITY-WEIGHTED CHECKS · MAX 142 PTS
SOUND ≥85% · CONDITIONAL 60–84% · INADEQUATE <60% · ANY FAILED CRITICAL CHECK FLAGS ITS LAYER
THE STANDARD: THE CONTAINED THING MUST NEVER ENFORCE ITS OWN CONTAINMENT