A capability constraint is a set of “restrictions on the use of a capability, including safety, context limitations, or policy-bound execution gates”. It is declared as CapabilityConstraint in the Agent Ontology’s capability module.1

An Agent Capability attaches constraints through hasConstraint. The constraint class has three string properties: allowedContext, the contexts in which execution is permitted; prohibitedContext; and requiresPermission. visionGraph places the class under Constraint and as a part of Agent Capability.1

A capability constraint limits what a capability may be used for. That is different from the Delegation Scope, which limits what one agent has authorised another to do, and from the Execution Constraint that applies to a particular delegated execution.

Footnotes

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