Great illustrations, Jason. From the components, I am thinking that the selector agent is the only 'true' agent as that is the part where autonomous decision making takes place (with an internal chain of thought). The other agents in the diagram appear to be just a large language model at work, doing singular specific tasks. What do you think?