The cooperative fault-tolerant for the nonlinear multi-agent system with hybrid actuator faults and multiple unknown control directions were investigated, based on the RAFSMT(robust adaptive fuzzy sliding mode technology). The piecewise Nussbaum function was used to solve the multiple unknown control directions, and the robust adaptive fuzzy technology was used to solve the nonlinear uncertainty in the system. A first-order sliding mode differentiator was introduced, and it was combined with the adaptive backstepping technique to obtain the first-order derivative of the virtual control law. At the same time, a robust bounded method was adopted to improve the convergence speed and tracking accuracy of the designed adaptive cooperative controller. Based on the algebraic graph theory, the error model of the agent system was established, the distributed consensus cooperative fault-tolerant controller was designed, and the Lyapunov candidate function was constructed to prove that the proposed controller can make the system stable. The effectiveness of the proposed method is verified by the comparison of simulation examples, which provides theoretical support for engineering practice.