CO is not needed for stability; however, it does improve transient response. Values of less than 0.1 uF could cause instability.
Maybe it's miswired?
If that's how it is actually wired, I don't see what the problem is.VCC2 can be 5V also, why not simplify as a test and ditch the 6V regulator?
The other 2 regulators are working completely find without external caps
but I'm completely confused as to how it fixed the issue