Remote monitoring and management (RMM) servers are skeleton keys: compromise one and you can push code to every endpoint it manages. The recent N-central zero-day, exploited hours after its proof of concept, is a reminder.
Questions to answer for your environment
- Is the RMM console reachable from the internet? Can it be put behind a VPN or identity-aware proxy?
- Which legacy features or APIs are enabled but unused?
- Do you collect process, network and authentication logs from the RMM server itself, not just the agents?
- Would you notice the RMM pushing a script to all endpoints at 3 am?
Hunting ideas
- New administrator accounts or API tokens on the RMM.
- Scripts or jobs deployed outside change windows, or to unusually many devices.
- Server processes spawning shells or download tools (see the shell-spawning rule thread).
- Outbound connections from the RMM server to unfamiliar hosts.
For MSPs: how do you separate customer tenants, and how quickly could you revoke RMM access to every customer if the server were compromised?