An agent task is a well-defined action unit exchanged between agents, declared as Task in the core module of the Agent Ontology.1 The threat-model module restates it and notes that it corresponds to the task graph.2 visionGraph places it under Task Planning.

An agent performs a task through performsTask, and an agent can hand one on with delegatesTask. A task requires capabilities through requiresCapability, whose range is Agent Capability; this differs from an agent having capabilities through hasCapability, and the two relations are kept apart. A task points to subtasks through hasSubtask, which expresses hierarchical Task Decomposition.

Task structure as a dependency graph is carried by the separate classes Task Node and TaskDependency. A task handed from one agent to another is a Task Delegation. An Agent Intent can be aligned with the task it authorises or motivates through IntentAlignsWithTask.3

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

  3. 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 ↩