A novel robust time-varying chain multi-agent control system is obtained. To solve the problem, control actions are formed in each agent of the network on the basis of the auxiliary loop method taking into account the information about the preceding agent. The signal from the leading subsystem is received only in the first agent of the network. The connection is one-way. The control system in the agent is built using measured data on the outputs of the agent itself and the agent preceding it. Compensation of perturbations in each agent of the network is realized by forming a special signal carrying information about all perturbations acting on the agent, and then its damping by means of an auxiliary loop. The construction of such a control system requires information about the derivatives of the intermediate signals of the system, for which two Halil observers are used, in a special way. To illustrate the performance of the proposed chain network system, we consider a numerical example of control a multi-agent plant consisting of six agents whose dynamic processes are described by nonstationary equations. The agents are subject to the action of uncontrolled external disturbances. Modeling in MATLAB Simulink has been carried out. The simulation results confirmed the theoretical conclusions and showed good performance of the chain system under conditions of uncertainty and nonstationarity of the network agents’ models.