do a first check of URL suitability (with something like the app's regex) and just alert the user directly that the page won't work with tinyThom.as rather than pinging the server to learn the same info
if the page's URL corresponds to search results, but there was only one result, the page doesn't have a sharable URL that tinyThom.as can even load. In this case, perhaps the javascript can grab the page's content and send it along, and we can parse that.
it would be nice if the bookmarklet's operation were a POST, not a GET
Bookmarklet wish list