Operator
Base class for entities that submit TaskPlans for execution.
| Declared in | src/libs/libopal/opal/Operator.hpp |
| Group | libopal |
Detailed description
Three operator types exist: - Sentient: Fully trusted human (or AGI). No permission prompts. - Directive: Trigger pre-approved by a sentient. No permission prompts. - LLM: Untrusted operator. Permission checks required.
LLM operators carry a shared PermissionStore that holds grants.
See also OpalTaskRuntime, TaskPlan, PermissionStore.