Skip to main content

Three RMM Incidents in One Week: Governing Extensions and Hosted Management Planes

5 min read
CISO Daily
Three RMM Incidents in One Week: Governing Extensions and Hosted Management Planes

The decision this week

In the span of about a week, three separate reports from Huntress and N-able touched remote management tooling. They differ in severity and in what is proven, so they should not be treated as one event. They do share a question for the board: which of your management consoles, and which add-ons to them, can be reached from the internet, and who owns that list?

What was reported

1. SCKit and ScreenConnect extensions. Per Huntress, on 2026-11-11 an operator used access to multiple cloud-hosted ScreenConnect panels to push commands that installed attacker-controlled ScreenConnect clients and a toolkit on at least 288 hosts across 48 organisations in about 90 minutes. A second wave on 2026-11-13 touched 205 hosts across 41 organisations. Huntress says the activity appeared related to RMM+, a third-party extension, whose versions 1.0.31 and 1.0.32 accepted a blank access key. ConnectWise disabled five third-party extensions across 400 partner environments. Huntress says most RMM+ installations it saw were legitimate, and that it has not seen console takeover or spread to technician workstations succeed.

2. N-able N-central (“Able2Shell”). Per Huntress’s updates, on 2026-11-09 a researcher published 33 vulnerabilities on GitHub and N-able took its hosted N-central offline. A hotfix build (2026.4.1.26) followed on 2026-11-10 covering CVE-2026-125878 through CVE-2026-125883 and CVE-2026-91000. On 2026-11-11 the researcher published two further proofs of concept; Huntress reports N-able says they need the non-default Legacy API to be enabled and internet-exposed, and that hosted servers are not vulnerable. In the 2026-11-10 update, Huntress relays N-able as reporting scanning activity associated with the published proof-of-concept code and no evidence that any customer environment had been successfully exploited. In its 2026-11-12 update Huntress said it had not observed exploitation. On 2026-11-13 Huntress reported that N-able had identified indicators of compromise for customers who have not disabled the Legacy API.

3. Fake sessions on exposed ScreenConnect servers. Huntress counted 2,408 unauthorized guest sessions under more than 1,000 fake names across over 100 servers, with 97.8% of connections from one address. It saw no malicious activity from those sessions. Its leading hypothesis is that they are bait for a ScreenConnect client flaw (CVE-2026-84869, CVSS 9.9 per ConnectWise, fixed in 26.6.5; we did not pull NVD or KEV for this CVE). Huntress presents that as a hypothesis, not an observation.

Our assessment

This is our reading, not Huntress’s.

  • On Huntress’s analysis, the weak point in the first case appears to have been an add-on, not the core product. ConnectWise’s advisory said it had not yet identified an initial access vector. If your risk register lists “ScreenConnect” but not its extensions, the register misses the component that mattered here.
  • A hosted console is a concentration of access. One panel reaches every endpoint beneath it. That makes the console’s access controls and audit logs a tier-one asset.
  • Disclosure can outrun vendor processes. In the N-able case the researcher published on 2026-11-09 and the hotfix followed on 2026-11-10. Start your response from the disclosure, not from the advisory.
  • Not every report equals compromise. Cases 2 and 3 describe exposure and probing, not confirmed victims. Calibrating that distinction keeps the board’s reaction proportionate.

Questions to put to your IT and security leads

  1. Do we have a complete inventory of RMM, remote-support and PSA tools, including each extension or plugin installed on them, with an owner for each?
  2. Which management interfaces are reachable from the internet? Does each have MFA, IP restriction and alerting on new sessions?
  3. Do we or our MSP use ScreenConnect, RMM+ or N-able N-central? If so, were versions and extension status confirmed against the vendor advisories this week?
  4. Can we pull the Run Command, Toolbox and audit history for our consoles, and has anyone reviewed it for activity nobody on our side initiated?
  5. If a technician’s workstation were compromised, what could it reach? Is that account separated from email and browsing?
  6. If a provider manages us, what does our contract say about notifying us of tool compromises and extension use?

Actions for the week

  • Ask MSPs and internal teams to confirm in writing: product versions, extension inventory, and whether any ScreenConnect extension was disabled by the vendor.
  • Request a review of panel audit logs for the period from 2026-11-11, and a sweep for unexpected second remote-access clients, using the indicators in Huntress’s posts.
  • Remove internet exposure of management interfaces that do not need it. For N-central, confirm the Legacy API status.
  • Add vendor extensions to third-party risk reviews, with a rule that an unowned extension is removed.

What we could not establish

  • As of 2026-11-16 we found no independent confirmation, in the sources we opened, of the SCKit initial access path beyond Huntress’s and ConnectWise’s statements.
  • The motive for the fake sessions is unknown.
  • The N-able statements about exploitation and the Legacy API are as relayed by Huntress; we did not open an N-able advisory.
  • Our assessment section reflects our judgement.

Sources