for testing LLM Tools / Agent
25-03-03:
- Update MessageState to Pydantic Format CleanDirectoryState.
- Edit @tool planner args to use Command with InjectedState and InjectedToolCallID.
- Edit @tool code_executor args plans(str) -> plans(Annotated[str, InjectedState("plans")]) (WIP) -> Need to edit @tool code_executor to write code -> test -> execute with HumanInterrupt