This work explores the development and implementation of a silicone foam soft sensor used in input devices for spatial computing, targeting low-effort, comfortable, compact, and discreet interfaces for interacting within three-dimensional environments. The soft sensor is composed of thin foam encapsulating a small magnet, and the deformation of the foam is monitored by a Hall effect sensor. Sensor characterizations quantified the sub-millimeter spatial sensing capabilities achieving a spatial resolution of 0.56 mm. Moreover, the sensor was scaled to three input devices, namely, a soft three-dimensional (3D) joystick for spatial control, a soft grip controller for multipoint grip input, and a force myography (FMG) wristband for hand gesture recognition. We evaluated the application of the sensor in a user study of the soft FMG wristband, where displacement of the wrist tendons were measured while users performed five typical hand gestures. Machine learning models were fit to quantify the accuracy of gesture recognition using the wristband. Our findings demonstrate the potential of the sensor used in input devices to enhance user interaction within spatial computing environments.