(Maybe intentional, maybe unintentional) deceptive advisories 101: https://certvde.com/en/advisories/VDE-2025-052/ .
The actual vulns here are OS command injection issues (CWE-78). The webapp just so happens to be vulnerable to CSRF too, so they use CWE-352, but honestly nobody in their right mind gives a crap about CSRF as a top priority item.
There are multiple ways to exploit the bugs. The score/vector in the advisory is technically correct, but you could also exploit the bug (or series of bugs) as 9.1 (AV:N/AC:L/PR:H/UI:N/S:C/C:H/I:H/A:H) or 9.9 (AV:N/AC:L/PR:H/UI:N/S:C/C:H/I:H/A:H) depending on the privilege required to do the OS command injection. But honestly the 'what privilege required' becomes moot when you search the user manual for default credentials....
Whether the deception is intentional or not, who knows, it is what it is. Attackers are never* gonna use CSRF, but they are absolutely positively going to abuse command injection (even authenticated command injection), especially against devices which has 1) a cellular modem and 2) published default credentials that are incredibly easy to learn.
This is all an example of the fact that CVSS does not score a vulnerability, but rather scores one exploitation method of a vulnerability. There are often multiple ways to interpret 'a vuln'. In this case the advisory probably should have reserved more CVEs anyway: some to cover the CSRF, and others to cover the command injection bugs (the fixes for each are most likely distinct code changes, so worthy of independent CVEs, but I digress).
And sorry for the sales pitch: this is the kind of thing that we manually review all week, every week, and publish details about in our Worldview reports: https://www.dragos.com/dragos-worldview/