Parts: coupon_h05 (0.5 mm living-hinge necks, one 0.4 mm line: the reference geometry) and coupon_h08 (0.8 mm necks, two lines: the fallback). Everything else is identical: a rectangular frame whose side members are the component's anchors, a bar that runs under a tunnel through each rail so the shuttle can be pressed from either side, and finger pads on both ends of the bar.

The frame is one closed loop. The first print (2026-09-10) failed because the plunger slots cut each rail clean through, leaving two separate side brackets held together only by the living hinges. Now the bar is 3 mm tall on the bed and each rail keeps a 2.75 mm bridge over it: the tunnel is open to the bed (so the bar prints in place without support) and closed on top, 5.2 mm of bridging per rail. The pads and the shuttle stay full height.

Print flat as built (bed at z=0, 6.35 mm tall), beams and necks in the bed plane. Enable thin-wall / Arachne perimeters for the 0.5 mm part. Print one of each in PETG first; the reference authors recommend polypropylene and PLA is expected to creep. The test: press a pad, does it snap and stay; press the other, does it come back; after 50 cycles, are the necks whitening. Record the result on the review page as a print report so lib/validation.json gains the evidence and the component's material_notes can be set.

Assumptions: span 37 mm and pre-tilt 8 deg are estimated from the reference renders (the coupon tests them); plunger and pads are sized for a finger, not for any hardware. Idea: ideas/bistable_compliant_switch.