An agent action is a concrete execution event initiated by an agent, producing observable or verifiable consequences.1 visionGraph places it under Event: it depends on an Execution Context and produces an Agent Artefact.

An agent performs its actions through executesAction. The core module connects an action to the artefact it produces (producesArtifact), to the Agent Trace Event that records it (recordedIn), and to the execution context that applies to it (contextOf). The intent module adds hasActionType to an Action Type, executedIn to an execution context, and loggedAs to a Runtime Ledger Event. An Agent Intent reaches an action through producesAction.2

The schema.org mapping aligns the class with schema:Action.3 An action is an event that has been initiated, so a plan or a requested step is an Agent Task or an Agent Intent rather than an action.

Footnotes

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

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

  3. Semantic Agent Communication Community Group, “Agent Ontology to Schema.org Mapping” (Agent Ontology, CC BY 4.0). https://github.com/s-agent-comm/agent-ontology/blob/750e06d276dd568313a6c03c4e78203887d3fa4b/ontologies/schema-org-mapping.ttl ↩