"Remote shell" for a pentest? I wouldn't bother with it. In a penetration testing gig your role is using to validate the ability to get into a system. While you may make mention of installation of software in your tactical plan to your client, what purpose would it serve. If you already got into a machine, you've done your job especially if you can install software on the machine so updloading Monad (powershell) unto a machine is pointless. It's kind of like saying "I broke into a bank and once there I decided to change the camera system to show X..." Eliminate all but the first five words... "I broke into a bank" your role is done. Installing software on a machine you've pentested can potentially break something somewhere down the line especially with Windows systems where reliance on DLL files are high.

True indeed that pentesting is more than just one machine, but you've missed the gist of what I was trying to get at. In most instances I try to avoid leaving footprints via way of software installs and even tool usage since many tools are pretty noisy. I've found that with a minimalist set of tools one can still accomplish most tasks. E.g. using telnet + sleep as opposed to always using nmap.

In using tools already in place, you're less likely to set off alarms or get tripped up if an admin (while doing say a zero knowledge) sees you and cuts you off. I personally like to enumerate and send targeted attacks as opposed to scripted ones to confirm/validate something. This is something automation is not good at (finding discrepancies). To each their own though. I would not install software if I didn't have to especially for the sake of automating the testing.