I’m subscribed to a lot of sites that uses external libraries to generate RSS feeds and ignore them (because “who uses rss this days?!”).
It would be great if I could specify XPath to extract single element from the webpage and display it inline. It could be totally client sided, i.e. loading page on the client and then drilling down.
I think it would improve usability by a lot, while not taking a lot of space to implement (as I think it could be basically configuration field of a feed).