I know this is against the main point of Anscombe’s Quartet but just curious: Could skewness or other summary statistics differentiate the four distributions?

Take enough moments and you'll be able to differentiate any distributions.