An operator with access to hosted ScreenConnect panels pushed commands to managed endpoints on 11 November that installed attacker-controlled ScreenConnect clients and a toolkit called SCKit. Huntress counted at least 288 affected hosts across 48 organizations that day and 205 more hosts across 41 organizations in a second wave on 13 November.
What Huntress and Blackpoint observed
According to Huntress’s analysis, suspicious activity began at about 17:47 UTC on 11 November across endpoints tied to multiple ScreenConnect environments. All 288 hosts were hit within 90 minutes, about a third of them in the first 10 minutes. The operator used the console’s Run Command feature to send commands to many endpoints at once, running as SYSTEM, which installed a second, unauthorized ScreenConnect client alongside the legitimate one and deployed SCKit.
Huntress recorded a second spike at 16:12 UTC on 13 November from new infrastructure. In that wave it saw rogue clients installed and repeated attempts to start the SCKit decoder script, with no follow-on executions observed. Installer requests in the campaign carried a counter in the form “Exploited” plus a number, which Huntress reads as a tally of hosts compromised across the campaign; one panel alone accounted for 190 downstream clients.
Blackpoint Cyber reported the same activity from its own security operations center, which began responding to unauthorized ScreenConnect deployments from legitimate cloud-hosted instances early in the afternoon of 11 November. In an update on 13 November, Blackpoint said it had seen a second wave that appears to target on-premises ScreenConnect instances where mitigations are still pending.
ConnectWise’s response and the RMM+ extension
ConnectWise, which owns ScreenConnect, posted an advisory at 22:45 UTC on 11 November. Huntress reports that the advisory named five independently developed third-party extensions that may be associated with the activity and said ConnectWise had not yet identified the initial access vector or the full scope. ConnectWise disabled the five extensions across more than 400 partner environments as a precaution. The list covers Labtech Plus, RMM Plus, RMM+, RMM Plus Password Link and RMM+ Password Link.
Huntress later linked the activity to RMM+, an extension that lets another tool manage ScreenConnect sessions without a technician working through the console. Huntress says RMM+ authenticates requests with a shared access key rather than ScreenConnect’s user accounts, roles and MFA, and that in versions 1.0.31 and 1.0.32 the key defaulted to blank, so the check passed when an administrator had never set one. Huntress found no login activity matching the commands, and the commands it reviewed appeared in logs linked to RMM+. It also says most RMM+ installations it encountered were legitimate, usually added for a project and left in place.
Blackpoint’s researchers were more cautious on this point. Their analysis found no technical link between the extensions and SCKit’s later stages, and treats the extensions’ possible role in the first panel access as unresolved.
The spread SCKit was built for, and where it stopped
Both firms reconstructed a second capability from the toolkit’s code. It is designed to wait for a technician to open a support session to an infected endpoint, push a file back to the technician’s workstation, take the technician’s authenticated ScreenConnect web session from that workstation, and then use the technician’s panel permissions to reach every endpoint that panel can manage. In an MSP, one technician’s panel can span many customers.
That step depends on CVE-2026-84869, a client-side ScreenConnect flaw rated 9.9 on the CVSS scale. ConnectWise fixed it in ScreenConnect 26.6.5, released on 8 September. Huntress describes it as a flaw in versions before 26.6.5 that “can allow an attacker to transfer and execute files through an active session without host authorization or confirmation.”
Neither firm saw the chain complete. Huntress says it has not observed a successful jump from an infected endpoint to a technician’s workstation and that “The console-takeover attempt failed in every recovered host log we have checked so far.” Blackpoint tested the recovered toolkit against a current ScreenConnect Host client: the malicious Guest queued the transfer, and the Host never requested or ran the file. Blackpoint says that result shows no bypass of the 26.6.5 fix.
The Huntress write-up, by researchers Ben Nahorney, Susannah Matt and Lindsey O’Donnell-Welch, puts one line at the top of its advice: “Defenders are advised to treat any ScreenConnect connection to an unfamiliar machine as unsafe.”
What it means for the security leader
The confirmed damage came from legitimate administration features run by someone who should not have had them. Run Command did what it was built to do. An extension added for a past integration, with a blank key, gave a caller a path that skipped the accounts, roles and MFA the security team had configured. RMM extensions belong in the same review as any other privileged integration: who installed them, what they can call, how they authenticate, and whether anyone still uses them.
The second lesson is about the support process itself. The standard reaction to an alert on a managed endpoint is to connect to it. SCKit was designed to turn that reaction into the next infection, and both firms tell responders to stop doing it until a host is cleared. Teams that run remote support for others, internal IT as much as MSPs, need an out-of-band route ready before an incident.
This is the latest in a run of attacks on the tools that manage other machines. We reported in September that a new ransomware group was weaponizing RMM tools, and N-able has shipped repeated fixes for N-central, including a fourth hotfix in weeks. ConnectWise also suspended a cloud backup service in October after unauthorized access to application identities, a separate incident.
What defenders should do now
- Disable the five extensions on any on-premises ScreenConnect server, as ConnectWise advises, and confirm they are off on hosted instances.
- Update every ScreenConnect client, technician Host clients included, to 26.6.5 or later. Blackpoint’s guidance names build 26.6.5.9742 or later.
- Audit Run Command and Shared Toolbox history for commands or tools nobody on the team initiated, and check managed endpoints for a second ScreenConnect client pointing at infrastructure you do not run.
- Do not connect to a suspected host through ScreenConnect. Huntress recommends WinRM or another out-of-band tool, and says to treat any technician workstation that already connected to a suspected host as potentially exposed.
- Revoke sessions from a clean device: end active ScreenConnect web sessions, rotate ScreenConnect credentials, and revoke browser and identity-provider sessions that may have been exposed.
- Keep affected hosts isolated until cleanup is verified. Huntress says isolation alone does not remove the persistence and that some systems were reinfected shortly after isolation was lifted.
Both firms published indicators of compromise. Huntress notes that file hashes are weak indicators here because the scripts are rebuilt with different embedded values on each host.