@Jeremiah from the

running-tasks.mdx

file in the repo i was looking at the manual execution of the agentic loop. i don't seem to understand a few things if you could clarify 1. why

orchestrator.run()

is being used here as the

cf.run()

is being executed at the agent level? 2. does a runtime status loop like

while task.is_complete()

instantiate a flow context within the loop? 3. does the

orchestrator

need to be defined before it can be called with the run method? 4. What is the purpose of explicitly calling the orchestrator run and what are the use cases? Almost all examples explicitly create a flow or flow context manager but this one doesn't