Most operations don't follow any "overwrite" concept.
To emulate it, one needs to check for existence of a resource, conditionally delete it, and then create the new one.
This is not easily applicable in a batch.

Implementing overwrite for all actions would be tedious.
But this could help:
Another "standard" property (similar to address and op-name) "onFail=ignore" or such, which would make the operation not fail if an exception occurs. A short CLI syntax could be a '?' at the end of operation's name:

/foo=bar/boo=baz:remove?()

This would allow mgmt API users to queue a removing operation without knowing if it exists; thus, effectively, doing an overwrite.