+1 on that, I use uniplate for pretty much all my haskell-src-exts tasks these days, works like a charm! I'd love to include some standard traversal functionality in haskell-src-exts that depends on uniplate, but hesitate to do so because of HP aspirations for haskell-src-exts. Neil, what do you reckon the chances of getting uniplate in the HP are? :-)
Or, you could just put a note in the documentation suggesting that uniplate is a good way to traverse parsed output, along with an example or two. And maybe mention a separate package haskell-src-exts-uniplate with utility functions, if there are any. Keeping deps under control by splitting packages this way has some precedent, and makes some logical sense too, since it's not like the parsed data types are trying to be abstract by hiding constructors.