When building more complex, longer-running agents, one of the main things that slows down development is needing to fully replay the workflow to test after modifying one node.
It would be cool if there were a way to test specific nodes either with mock input data or by using the inputs from the previous run.
