you should ensure that that he agent block is instead a task block like the image attached whereby each task has a drop down with details of subagents being used to execute that task, tools they are using and mcps they are using so that the user can see the stream of actions being taken by the agents to execute the task at hand
ok you nailed it however, it shouldn't be an agent block it should be a task block, look at the image for what i mean where each task is also a drop down with details of that task being executed by agents in a streeam of thinking plus using tools, mcps and subagents
no still not quite there, you didn;t change the task pane to be more like i want so that the user can inpsect the claude like working of thinking, planning, executing tasks, and in each task the ability for a drop down which will fill with streamed data of subagents, tools, and mcps being used and what was found, this is a multi agentic hierachal agentic system which mirror much of claude code so think about what that can do , in addition to mirror what cursor can do but for docs
I like your thinking however, iw would like it so that the user can see each task being executed and see in a drop down for each task waht is being done by the central brain in terms of tools used, subagents, and thinking behind its process so that they can inspect if they want
ok please add in a task pane which replicates cursors task plan whereby it has plan, to do list creates, spinners that spin as each task is being executed which can be dropped down to see the agents working, and then a final summary under the task pane of what it achieves, this should be in the chat side bar, so it goes user message which is complex ask, then this task pane appears in clean modern cool looking, and then once finished has summary under task pane of what it did
