[Request] The model of agents in LobeHub agent team should support adjusting reasoning depth and output verbosity #12776