I'm a day or two away from "finishing" a script that will allow you to do that (well, at least as far as Google will let me) and pushing it to Github. I'm likely not the only person working on this so I expect we'll be able to save most of our data.

It can be done but it isn't pretty.

My first resort when someone deletes their blog or whatever thing that was meaningful to me that I foolishly failed to mirror, is to go to archive.org and see if they have a good mirror. If they don't but they do have enough that you can find out the RSS URL, you can get google reader's archive of the feed like so: http://webapps.stackexchange.com/questions/17484/archive-exp...

What you get back is just raw XML.

EDIT: for example, here are the last 100 posts from Mark Pilgrim's blog: http://www.google.com/reader/atom/feed/http://diveintomark.o... ... in this particular case archive.org has a good copy though so this isn't necessary as a last resort.

Closest thing you can do is grab the JSON of the data as the browser downloads it. The cached content that shows up on a feed's pane is transferred in bulk via JSON, so it should make a decent archive.

Just need a way to do it automatically...

I pushed a small python script to allow you to export your read items as JSON here https://github.com/motdiem/GRARchiver - this can help