Setting up a splitter (which I just tried to do, so my memory is fresh) goes like this:

1. Somehow manage to get the beam to the screen at the end.
2. Slowly turn on quads, a bit at at a time, trying to improve transmission, or at least maintain it. Use the screen for this.
3. Try to get the beam to be centered in the quads, but this can be very hard at first, when transmission is low, and the beam is huge and sampling lots of nonlinear fields.
4. Get to near the design quad currents, choose currents as close as possible, while still maintaining a "sensible" beam on the screen.
5. Do final quad alignment.
6. Correct dispersion / R56.

Step 6 is the only automated step, and the day-to-day tune up is almost entirely 5&6, so your script will do most of the rest. We need to devote some time to test it!