This.

The ability to remotely run arbitrary code on a machine that intentionally gives SSH shell and write+execute access to the filesystem is not a vulnerability just because it's a productivity aid to users who want to leverage this access to do bad things on the remote machine.

A remote access protocol that gives a potentially untrustworthy remote system the ability to execute arbitrary code on the local machine is a serious problem in any scenario where the remote connection is presumed to be a one-way trust boundary.

Which of course includes any scenario where I myself deliberately run untrustworthy code on the remote, no matter how much I trust the remote itself and its owners.