An agent plans the tool calls, shows you each one before it runs, and keeps the results in one place.
No experiments yet. Create one, then give the agent a research objective.