Read/WriteWeb has a post on Wine.com and it's RSS API. You can see an example of the API (i.e. a REST GET) here; you can see how well it validates here (not very, though the fixes look to be very simple).

Every page (well, except admin stuff) in BlogMatrix is an "RSS API" in the same sense as this. Our production model is quite simple: a userid + a set of tag selectors + a template selector = output. The userid is encoded the domain name; "all" acts as a wildcard and our new "network" feature acts as a wildcard across a defined set acconts. The tag selector is defined by the path of the URI: for example, this selects all posts tagged "blogmatrix" and "downtime". Finally the template is selected in one of three ways: ending a path in "index.xxx" selects template "xxx"; "?t=yyy" selects template "yyy" (and will override index.xxx if present). If neither of these are there, the query is assumed to be "index.html" which selects the template "html". Thus, to convert the page above to an RSS feed, just add "index.xml". If there are structured data elements in the posts they'll be properly encoded into the RSS feeds.

The challenge for the BlogMatrix Platform is to take an RSS feed like Wine.com's and when it is imported to store it in a useful way. We're 95% of the way there, but it's not a priority for short term delivery. Still...