This paper presents a safety-constrained distributed sliding-mode control (SMC) framework for vehicle platoons subject to nonlinear dynamics and time-varying, possibly asymmetric constraints. The approach couples (i) a constraint-aware reference generator with Hermite smoothing, (ii) an improved prescribed-performance function (PPF) that uses a singularity-free error transformation with adaptive boundaries, and (iii) a topology-respecting three-layer SMC with a continuous boundary-layer reaching law. Lyapunov analysis proves finite-time reaching and asymptotic stability, and shows that tracking errors remain within prescribed bounds and that the control input is bounded with reduced chattering. Simulations across four information-flow topologies demonstrate improved performance across all considered cases.