This probably won't be of much interest to others, since it's relatively easy to interrogate other sites for moon data (e.g. moon lit %, the side that's lit, moon image, moon age), but if you don't care too much about being very precise and only want to base your skin output strictly on Weather.com's wxdata values (without interrogating other sites), you can still approximate the moon lit % based on the moon icon value (that is actually supposed to mean moon's age in days since the last new moon). At first I set these hardcoded, taking a more symmetrical (in terms of moon lit % increase/decrease ratio) synodic month data from NASA, but the formula for a rough approximation is quite simple:
As you can see, the graphs are pretty much the same in terms of increase/decrease ratios of moon lit %. Of course, the values will vary depending on the synodic month in question and latitude, but it's a good reference for approximations. The Calc measures where these formulas will be used will output either the moon lit % approximation (the 1st formula), or the side that's lit (the 2nd formula, that outputs -1 for left side lit, 0 for bottom lit and 1 for right side lit).
Note: And now, of course, we can see why wxdata's moon icon value is supposed to represent 2 full moons one after another, for 14 and 15...