I've managed to go one big step better than my first round of information:
http://philhart.com/cloud_forecasts
It's possible to produce cloud forecast maps like the one below. This forecast was made yesterday for midday today (compare it to the satellite image for the same time).
Having a map enables much better intepretation than just boiling cloud cover down to a single percentage figure for one location.
I've also put a page together with a set of images comparing an actual infrared satellite image with forecasts made on the same day and up to 5 days ahead. The upshot is that forecasts 2-3 days ahead are pretty good! Beyond that it gets a bit flaky..
http://philhart.com/cloud_forecast_accuracy
cheers
Phil