Skip to main content

Command Shell behavior change in 25.3.1 – security and policy control concerns

  • March 20, 2026
  • 2 replies
  • 18 views

Forum|alt.badge.img+5

Hi everyone,

we would like to raise a concern regarding the Command Shell behavior introduced in Remote Support 25.3.1.

The fact that the shell now starts in the context of the logged-in user by default (without requiring user credentials or consent) is, from our perspective, a significant security regression.

In previous versions, this behavior was not possible — access to the user context required explicit authentication. Now, support staff can directly access user-level resources such as OneDrive, network shares, or SharePoint without the user being aware.

In regulated environments (e.g., banking), this is a serious issue:
- it breaks the separation between user and system privileges
- it introduces potential audit and compliance risks
- it allows unintended access to sensitive user data

At the same time, our operational requirement remains:
- we must be able to perform support actions without user interaction
- but strictly within system/administrative context unless explicitly approved otherwise

Currently, we do not see any way to control or restrict this behavior via session policies or configuration, which is a major gap.

Questions to the community (and hopefully BeyondTrust team):
- How are you mitigating this risk today?
- Is there any undocumented workaround to restrict user-context shell access?
- Is BeyondTrust planning to introduce policy-level controls for this?

From our perspective, this is not just an enhancement request but a necessary control for secure operation in enterprise environments.

Without such control, we see a real risk that the product may not be compliant with internal security standards, which could impact its continued use.

Would appreciate feedback from both the community and BeyondTrust.

2 replies

Forum|alt.badge.img

I’m on 25.1.7 and can confirm the shell jump puts the tech in the system context.

I wonder if they did it to mitigate the opposite risk.  In previous versions, whether or not someone in your position was supposed to have admin rights to computers, the shell jump automatically gave you System level access - which in itself is a security issue of the highest kind.  That would mean that anyone with access to jump into a machine automatically had admin rights to the machine.  I could be wrong, but the only other context that it could automatically put itself into is the currently logged in user’s context.  Sounds like they traded one evil for a much lesser evil.

It would be great if they could provide the option of which type of shell access it had (system vs user context).  I would say this is looped into the RBAC we’ve been told will “be released very soon” for the last like 4 updates.


Forum|alt.badge.img+1
  • Veteran
  • March 20, 2026

We’re currently on Remote Support 24.2.3.

In our environment, Command Shell executes under the LocalSystem security context, but it can still access the logged-on user’s OneDrive (e.g., ‘C:\Users\{userID}\OneDrive - {Org Name}\Documents’). User-mapped drive letters are not available from the shell.

Also worth noting - when we start a GUI process from the Command Shell (e.g., notepad.exe), it is visible to, and interactive for, the logged-on user, even though the process is running as SYSTEM. This is one of the major reasons we try to avoid using Command Shell when possible, since it can still disrupt the end user’s session. After ending the Jump Session, the process persists for the user.