To address the mission requirements of spacecraft swarm covertly approaching and observing space targets, this study proposes a swarm covert maneuver trajectory planning method based on improved sequential convex programming (SCP). Given the numerous, nonlinear, and strongly coupled constraints in multi-satellite cooperative covert maneuvers, strategies such as polyhedral cones and rotating hyperplanes are employed to convexify constraints related to covert zones and collision avoidance, thereby reducing problem complexity and improving computational efficiency. Additionally, to prevent premature termination of the SCP solver due to improper initial guesses, a penalized iterative recovery mechanism is introduced. Simulation results demonstrate that the proposed convex approximation techniques effectively reduce problem-solving difficulty, and the improved SCP method successfully generates feasible swarm covert trajectories that satisfy complex constraints.