A rotating body of angular momentum J drags local inertial frames around with it (Lense-Thirring effect, 1918). In the weak-field, slow-rotation limit the frame-dragging angular velocity in the equatorial plane is ΩLT = 2GJ / (c² r³), with J = Iωspin and I = k M R² (k = 0.35 here, uniform-density-like guess).

The test gyroscope orbits with the Kepler rate Ωorb = √(GM/r³), so each orbit its axis turns by Δφ = ΩLT·Torb. That is tiny (about 0.1 degree per orbit for the default star), so the drawn axis uses a visual gain; the numbers in the readouts are real (gain 1).

Honest simplifications: Newtonian orbit, weak-field formula even though r = 3 R is not weak for a neutron star (real values differ by tens of percent, and frame dragging in the full Kerr metric also changes the orbit itself); a gyroscope precesses at GJ/(c²r³) for an equatorial orbit while 2GJ/(c²r³) is the dragging of the local frame; neutron-star radius fixed at 12 km; rigid spin, no magnetic field.

Gravity Probe B check: for a polar orbit the orbit-averaged gyroscope drift is GJ / (2 c² a³). With Earth J = 5.86e33 kg m²/s and a = 7027 km this gives about 41 mas/yr; the satellite measured 37.2 ± 7.2 mas/yr (GR predicts 39.2).