MARC: A Control and Uncertainty Disclosure Profile for Generative Models and Agents

This document specifies MARC, a vendor-neutral control and uncertainty-disclosure profile for generative models and agentic systems. MARC defines a small set of interoperable control metadata, separates pre-decision capability assessment from post-decision answer confidence, identifies the target of confidence disclosures, and defines a bounded primary action set for answering, clarification, retrieval, tool use, additional deliberation, abstention, and escalation. MARC does not standardize model internals, training methods, agent discovery, authorization, transport, tool schemas, provenance systems, or claims about machine cognition. Instead, it defines externally observable semantics that can be implemented by model providers, orchestration layers, evaluation harnesses, API gateways, and user-facing systems. The goal is to reduce silent failure, unnecessary externalization, and misleading uncertainty communication while improving auditability and interoperability.