Create a 15-second vertical practical-stunt action comedy titled GREEN MEANS GO. Fictional adult cast: one athletic woman in a mustard-yellow windbreaker and black trousers, and two adult pursuers in charcoal tracksuits. Preserve their faces, wardrobe and body proportions across cuts. One coherent quiet city block with a parked sedan, a low pedestrian railing and a red/green pedestrian crossing. No moving traffic. 0.0-2.8s: Begin on action, with the woman already vaulting across the parked sedan hood. Low 28mm side tracking camera follows her hand plant, body crossing the hood and both feet landing on the pavement. The two pursuers rush into view behind. Weight, hand contact and landing must read clearly. 2.8-5.8s: Hard cut to close rear tracking at running speed. She clears one waist-high railing in a single leap; the pursuers turn around its end. The camera moves with the runners and holds clear geography. 5.8-8.6s: Cut to a frontal 35mm medium-wide. She abruptly stops at the pedestrian crossing while its signal is red. Both pursuers skid to a stop beside her. All three are breathing hard. The threat collapses into awkward politeness. 8.6-10.6s: Quick close-up of their impatient glances, then the red pedestrian signal changing to green. Keep this short and readable. 10.6-15.0s: The crossing beeps and all three explode back into the chase at once. A fast lateral tracking camera follows their acceleration across the empty crossing; she leads into another leap as the shot ends on continuing motion. Light: hard warm afternoon sunlight from camera-left at 4300K, cool reflected building fill from camera-right at 6500K. Natural skin and fabric texture, clean directional shadows, grounded shoes and consistent contact shadows. Controlled handheld camera with readable faces and bodies. Diegetic audio: urgent footfalls, jacket movement, a sharp shoe squeak as they stop, breath, then the pedestrian-crossing beep and renewed running. Sparse original suspense percussion drops out during the red-light pause and returns on green. Keep the native generated audio. Aspect ratio 9:16, requested delivered format 720x1280. No captions, logos, celebrity likenesses, guns, military uniforms, blood, injury, teleporting, extra limbs or morphing vehicles.