I suggest two things:

1) Every single bug needs to either include a save that displays it clearly, or a method that shows it clearly. This is critical. It's not enough to say "it's been tested," you have to specify what the test is so CA can see for itself. There's no point otherwise.

2) Only the most egregious, consistent, and specific AI flaws (e.g., "loading a saved game severely inhibits AI aggressiveness in the first turn after the load") should be mentioned. AI is immensely complex, and you can't fix a problem like "the AI doesn't attack when its forces are clearly superior" the way you can fix a UI problem or an animation problem or whatever. There's no point in including fundamental AI inadequacies here—of course they know that the AI isn't top-notch. We don't need to tell them that.

-Simetrical