This tool tests whether a constrained blind-session protocol can generate target descriptions that can be scored after reveal without relying on open-ended interpretation. Visit tool ⇒
