graph LR
Agent_Orchestrator_Core_["Agent Orchestrator (Core)"]
Planner_Module["Planner Module"]
Executor_Module["Executor Module"]
Self_Refinement_Component["Self-Refinement Component"]
Agent_Orchestrator_Core_ -- "delegates tasks to" --> Planner_Module
Planner_Module -- "provides plans to" --> Agent_Orchestrator_Core_
Agent_Orchestrator_Core_ -- "delegates actions to" --> Executor_Module
Executor_Module -- "provides results to" --> Agent_Orchestrator_Core_
Agent_Orchestrator_Core_ -- "initiates" --> Self_Refinement_Component
Self_Refinement_Component -- "provides feedback to" --> Agent_Orchestrator_Core_
Self_Refinement_Component -- "requests replanning from" --> Planner_Module
Self_Refinement_Component -- "requests repair from" --> Executor_Module
The Agent Core (Orchestrator) subsystem is the central intelligence and control unit of the OS-Copilot project. It orchestrates the overall workflow, manages task execution, and drives the self-refinement process.
The central control unit responsible for initiating and coordinating the agent's operational flow, including planning, execution, and self-refinement cycles. It acts as the primary decision-maker and task delegator.
Related Classes/Methods:
Responsible for generating initial task plans, decomposing complex tasks, and revising existing plans based on feedback or new requirements from the orchestrator or self-refinement component.
Related Classes/Methods:
Facilitates the execution of planned actions, including generating necessary tools or code, and handling the repair of execution failures.
Related Classes/Methods:
Provides adaptive capabilities by evaluating outcomes, judging performance against objectives, and triggering corrective actions such as replanning or repairing execution issues.
Related Classes/Methods: