2 comments

[ 3.2 ms ] story [ 16.2 ms ] thread
Technical details here: https://grepular.com/Abusing_HTTP_Status_Codes_to_Expose_Pri...

The trick is to identify GET requests that will succeed only if the victim is logged into $SITE_OF_INTEREST, and bury them in an

   <img src="https://SITE_OF_INTEREST/more/stuff/here" 
        onload="is_logged_in()"
        onerror="not_logged_in()"
        ...>
If $SITE_OF_INTEREST doesn't have decent CSRF protection, this is an easy way for a rogue website to not only make a request, but observe the result.