[12:38] <smoser> pquerna, i want to solve the problem you are wanting to solve.
[12:39] <smoser> but the goal i think is to end up with 2 somewhat-commonly formed files with metadata information.
[12:39] <smoser> one would contain any thing that potentially was sensitive (ie, user-data) and one non-sensitive.
[12:39] <smoser> and their on-disk permissions represent that difference.
[12:40] <smoser> and then 'cloud-init query' would basically load those up and spit out answers, but you could just as easily load the json directly from /var/lib/cloud/instance/data.json 
[12:40] <smoser> which i guess would ideally be in /run.