URL autolinking appears to employ very crude patterrn matching to autoamgically determine what is a URL/external link - namely it ssems just to search for wwww at the beginning of a string. This causes it to auto link things that are not URLS as long as they match this criteria. www.domain.name.crt gets autolinked for example.
Better patch matching needs to takepalce to check if it is actually a valid url. I think there should also be a config option to turn off autolinking all together, or allow it to be escaped. Even valid URL's might not be alive for example, and I, for one, would prefer to chose what gets linked, and ther are good reasons **not** to link to some URLs, - links to web pages that launch exploits etc.