This is why I use a native sshd; one of its features is a customisable shell. (thanks Bitvise)
BTW, since PS itself runs as a bog-std Win32 console app (the difference: fancy colours/type when invoked by itself), I think it would be nothing more than setting up a shell script/batch file that spawns PS at startup...
This is PowerShell running from the Korn Shell/UWIN (by Southwestern Bell)
Which means that a shell script executed at startup, should be all that is needed to spawn PS from an OpenSSH session...
Well, I've gone and done it: running PS from an SSH session:
No, that is NOT fake; as PS1.0 is a .NET2.0 managed app, all that is needed to run it, is to patch the hotfix installer (modify the INF, then slip the original *signed* copies *while* CryptSvc is starting up), related files, and PowerShell.exe. (change subsystem version from 05 00 01 to 05 00 00)
IOW, managing a Windows .NET Nano server via SSH *may* yet be possible.