I'm sure others will comment but my 2p on this.
Firstly you could add the functionality yourself to Calibre if you wanted to and had the Python skills/time to do so. You could do a gui plugin that does add and convert operations exactly how you want them.
In terms of auto-conversion, the issue I see is what about configuring the options for conversion? These will vary by input format and "quality" of the format you are inputting from. If those are all under your control (say you had loads of text files you had created formatted with markdown text and setup your default options for text conversion to match) then I guess relying on default conversion options could work.
However (and this could just be me not getting fully to grips with Calibre's conversion options yet) I find that I am rarely satisfied with the EPUB that Calibre generates. Nearly every time I have to go to Sigil to sort out formatting, TOC, correct chapters etc to how I like them. Yes at least some of those can be done by Calibre configuration, at the expense of "fiddling" book by book with regular expressions, xpaths, line break settings etc etc. In some cases there are things I cannot yet find how to do what I want within Calibre (like specify paragraph margins/indents in fractional em instead of points and center h1-h4 headings). That is almost certainly just ignorance on my part, I haven't scoured that part of the manual yet.
My point being that in my own workflow I try to get a "how I like it" EPUB created first, and only then do I convert the EPUB to MOBI and delete the other formats. Sometimes the different formats of a book have come from different sources or conversions so one may convert to EPUB better than another. I guess I'm saying it is not a feature I would rely on to "automate" things for me, as wonderful as it would be to take all the manual eyeballing out!
Absolutely my scenario is not everyone's though. If you are less picky about what Calibre generates without per book customisation or you have better quality inputs to suit it then I can see it being wonderful to automate those steps.
[EDIT]Oh, one more "issue". You mention automating getting metadata with the add, again not unreasonable although it would dramatically slow down the add process. However what about the books for which metadata retrieval fails? You are then potentially doing a conversion with dud data like missing covers etc.
Last edited by kiwidude; 01-09-2011 at 08:23 PM.