Script Signing
Every PowerShell script ZenPrint runs on your endpoints (check-in, scan, reconcile, capture, poll) is cryptographically signed, and your endpoint verifies that signature before running anything. If a script’s contents have been altered, or it isn’t signed, the endpoint refuses to run it.
The important property: the signing key is not on the ZenPrint server. Scripts are signed ahead of time, and the server only ever hands out the already-signed copies. So even if our server, or the network between you and us, were compromised, no one could substitute a malicious script and have it run on your fleet, because they couldn’t produce a valid signature for it.
Everything is fetched over HTTPS as well, so signing sits on top of normal transport security as a second layer.
The right script, and not an old one
A valid signature shows that ZenPrint signed a script at some point. On its own it doesn’t show that the script is the one your endpoint asked for, or that it is current. So every script also states, inside the signed part, which mode it is for and a serial number that goes up when a script gets a security fix.
From bootstrap 1.7.0 your endpoint checks both after the signature:
- A script for a different mode than the one requested is refused.
- The endpoint remembers the highest serial it has run and refuses anything lower. An old signed script with a known problem can’t be served again, even by someone who controls the server.
- A script with no such statement is refused. That covers every script signed before this check existed.
Bootstraps older than 1.7.0 don’t do these checks. The dashboard shows the update banner for them; see Bootstrap for how to update.
What this doesn’t cover
Signing protects the scripts ZenPrint sends you. It can’t protect the bootstrap script you paste into your RMM. That’s the trust anchor, and if someone could swap it out inside your RMM, the checks downstream wouldn’t matter. Treat the bootstrap like any production credential: only download it over HTTPS from your ZenPrint portal, and control who can edit your RMM’s script library.
It also can’t protect against a compromise of the RMM agent itself, which runs whatever it’s told regardless of our checks.