Trusses are extensively used in large space structures—such as giant solar arrays and large-aperture telescopes—where assembly and maintenance often require robots to travel far from the central body. Bipedal climbing robots, benefiting from their flexible mobility and enjoying a larger reachable workspace, have distinct advantages for long-distance and confined-space tasks. Bipedal climbing robots must always maintain at least one gripper in contact with the structure to avoid falling, making their reachable configurations are inherently discrete, thus planning a safe, sequential grip strategy is critical. In this paper, a novel 7-DOF bipedal climbing robot is designed, equipped with dual grippers for efficient truss traversal. The movement of each gripper is decomposed into three stages: loosening gripper, moving and gripping structure. We introduce an A* based grip-sequence planner that generates a path along the truss from a given start pole to a target location, which selecting grip poses that respect safety margins and maximum step distances. The approach is validated in NVIDIA Isaac Sim, demonstrating effective path planning and smooth, robust climbing maneuvers across truss geometries.