It's an ad. I'll bite, though.

#1 The Internal Job: the real mistake was in having production credentials (i.e., the keys to the generators) inside the codebase. The keys should have been generated when the software was deployed. Hardcoded information should be kept to a minimum, or ideally eliminated altogether.

#2 Breaching the Version Control Process: there's not really enough information here to figure out the actual issue. Someone changed a version number, and that caused a bug to reappear? That sounds like #ifdefs or similar, where setting a version number can cause code to be included or excluded. Don't do that: tag your versions in your VCS, and eliminate dead code ruthlessly. If you really need to get it back, you can always use your VCS to restore a particular versions' code.

#3 Exposing from the Cloud: git's great; GitHub's great; not knowing to keep your repos private is…not so great. Also, pretty damned easy to avoid. Any private VCS would work here, to include git repos on a filesystem.

RhodeCode might be neat; it might not be. No doubt it could possibly provide for a lot of stuff out of the box. At the end of the day, though, you can't replace smart people executing well-designed processes with software; software is only good for replacing drones executing dumb processes.

So I agree having good version control tools is always a must but some common sense is always needed too. Although that said, I know code was reinserted on me once (similar to example 2) but there is nothing more frustrating than dealing with idiots.