A critical capability is a kind of Agent Capability whose exercise is consequential enough to need distinct handling. It is the sibling of Restricted Capability, and the two classes are disjoint. The Agent Ontology declares CriticalCapability without a definition; this one follows from its position and its one axiom.

The security-binding module declares it as a subclass of the core Capability with owl:disjointWith RestrictedCapability, under the heading that agents cannot claim conflicting capabilities.1 The threat-model module restates the axiom to prevent contradictory capability claims.2

No capability may therefore belong to both classes, and a reasoner flags an agent whose declared capabilities fall into both. Which capabilities count as critical is for the deploying party to decide.

Footnotes

  1. Semantic Agent Communication Community Group, “Security Binding Ontology” (Agent Ontology, CC BY 4.0). https://github.com/s-agent-comm/agent-ontology/blob/750e06d276dd568313a6c03c4e78203887d3fa4b/ontologies/security-binding.ttl ↩

  2. Semantic Agent Communication Community Group, “Threat Model Ontology” (Agent Ontology, CC BY 4.0). https://github.com/s-agent-comm/agent-ontology/blob/750e06d276dd568313a6c03c4e78203887d3fa4b/ontologies/threat-model.ttl ↩