TL;DR
The Qlik Sense reverse shell attack involved blocked PowerShell launch attempts, but the same attacker infrastructure exposed separate ActiveMQ and Veeam exploit tooling.
- Cynet observed suspicious
scheduler.exe process activity across several European customer environments in October 2026.
- Infrastructure analysis uncovered a custom Go backdoor, SSH persistence scripts, and exploit material targeting CVE-2023-46604 and CVE-2024-40711.
- Organizations should investigate suspicious process activity, review server exposure, and validate ActiveMQ and Veeam patch levels independently.
How Did the Qlik Sense Reverse Shell Attack Begin?
The Qlik Sense reverse shell attack involved suspicious PowerShell execution attempts against servers across several European customer environments in October 2026. Cynet’s researchers identified a recurring process pattern involving Qlik Sense’s scheduler.exe, followed by cmd.exe and PowerShell. However, Cynet blocked the script before the reverse shell process launched in the observed incidents. The exact vulnerability responsible for the initial Qlik Sense access remains unidentified.
Cynet identified an exposed directory containing Windows and Linux Go-based agents, SSH persistence scripts, and exploit-development artifacts. These included tools targeting Apache ActiveMQ’s CVE-2023-46604 and Veeam Backup & Replication’s CVE-2024-40711. The findings suggest preparation for additional intrusion opportunities, rather than evidence of successful exploitation of those products.
What Did the Qlik Sense Process Activity Reveal?
Cynet observed Qlik Sense’s scheduler.exe spawning cmd.exe, which subsequently launched PowerShell with hidden execution arguments. This resembled the process pattern documented during the November 2023 CACTUS-linked Qlik Sense intrusions. However, the resemblance does not establish common attribution or an identical exploit chain.
The decoded PowerShell launcher contained two operations. First, a WebClient request was configured to send the computer name and username to the attacker’s infrastructure. Next, a hidden PowerShell process would have attempted a reverse-shell connection. Cynet blocked the script before that process launched.
The reported destination was 45.115.26[.]23:4455. The investigation does not establish that an interactive reverse-shell connection was established on the affected customer systems.
Cynet subsequently tested the reverse shell in a controlled environment. The interactive communication channel supported host reconnaissance and delivery of persistence scripts.
The researchers also identified a Go-based Windows agent connecting to the same infrastructure over port 4456. Its Linux variants supported multiple architectures. Sandbox analysis revealed commands for system discovery, ActiveMQ process identification, and attempts to inspect sensitive Linux account files.
The associated persistence artifacts included OpenSSH installation scripts and a scheduled task named WinUpdCheck. One PowerShell script was configured to execute a reverse shell periodically from C:\ProgramData\ssh\winc.ps1. These findings demonstrate the toolkit’s capabilities, not confirmed persistence on the affected customer systems.
The exposed infrastructure contained separate exploit-development materials for two enterprise products. Cynet assessed that the operators appeared to be testing the ActiveMQ and Veeam exploits, although the recovered tools were not production-ready.
Apache ActiveMQ: CVE-2023-46604
Cynet recovered three XML files identified as ActiveMQ exploit material: poc_amq.xml, poc_amq2.xml, and poc_amq3.xml.
CVE-2023-46604 is a remote code execution vulnerability involving unsafe class instantiation during Java OpenWire processing. A manipulated OpenWire command can cause a vulnerable broker or client to instantiate classes from its classpath. Exploitation requires network access and a suitable class capable of executing code when instantiated with a string parameter.
Apache published fixed releases for its affected ActiveMQ branches, including:
- ActiveMQ 5.18.3
- ActiveMQ 5.17.6
- ActiveMQ 5.16.7
- ActiveMQ 5.15.16
- ActiveMQ 6.0.0
For ActiveMQ Artemis, the recommended upgrade was 2.31.2. Artemis contains the vulnerable OpenWire dependencies. However, it does not ship with Spring dependencies. Apache reported no known exploit against Artemis at the time.
Importantly, Cynet did not observe successful ActiveMQ exploitation in the investigated customer environments. The recovered files indicate exploit preparation, not confirmed compromise through this vulnerability.
Veeam Backup & Replication: CVE-2024-40711
The researchers also recovered a modified version of a public exploit toolchain targeting Veeam Backup & Replication.
CVE-2024-40711 is a deserialization vulnerability that can allow remote code execution. Veeam classifies CVE-2024-40711 as an unauthenticated RCE vulnerability. However, watchTowr identifies an authentication requirement for build 12.1.2.172. Builds 12.1.1.56 and earlier are identified as vulnerable to unauthenticated exploitation.
Cynet identified a modified tool based on ysoserial.net and a public watchTowr exploit implementation. The recovered source targeted Veeam’s Mount Service remoting endpoint on TCP port 6170.
The recovered files contained additional logging and debugging changes. Researchers also identified a loopback callback address in the serialized payload. These details were consistent with local testing rather than a demonstrated production exploit operation.
Veeam identifies Backup & Replication 12.1.2.172 and all earlier version 12 builds as affected by CVE-2024-40711. Unsupported versions were not tested but are likely affected and should be considered vulnerable. Veeam addressed the vulnerability in version 12.2, build 12.2.0.334. Organizations should verify their installed builds against Veeam’s official security bulletin.
The recovered Veeam artifacts indicate exploit testing, not confirmed exploitation during the October 2026 activity.
What Should Organizations Investigate After These Attempts?
Organizations should investigate Qlik Sense process lineage, suspicious outbound connections, persistence artifacts, and ActiveMQ and Veeam patch status.
Prioritize the following actions:
1. Review Qlik Sense process lineage. Investigate unexpected scheduler.exe child processes, particularly cmd.exe and hidden or encoded PowerShell execution.
2. Check outbound connections. Review historical network telemetry for connections involving 45.115.26[.]23, particularly TCP ports 4455, 4456, and 4457. Also investigate HTTP access to port 8012, which hosted the exposed staging directory. Treat these as reported indicators requiring contextual validation.
3. Investigate persistence artifacts. Search for unexpected WinUpdCheck scheduled tasks, suspicious PowerShell files, and unauthorized OpenSSH service installations.
4. Validate ActiveMQ versions. Identify affected brokers and Java-based OpenWire clients. Apply Apache’s branch-specific security updates.
5. Verify Veeam Backup & Replication builds. Confirm that affected installations have moved to version 12.2.0.334 or an appropriate later release.
6. Correlate endpoint and server telemetry. Check for unusual process creation, unexpected network connections, and suspicious file or service changes across related systems.
The priority is to determine whether the blocked Qlik Sense attempt was isolated or accompanied by separate persistence or exploitation activity elsewhere in the environment.
Hexnode XDR can support process-level investigation and containment on supported, enrolled Windows endpoints. Its process investigation capabilities are relevant to the suspicious parent-child execution pattern identified in the Qlik Sense activity.
The Incidents tab provides process metadata, including command-line arguments, file paths, integrity levels, and file hashes. The Process tab presents parent-child execution relationships, helping administrators investigate suspicious process chains.
For confirmed malicious activity, administrators can use the response actions:
- Kill Process: Terminate an individual suspicious process.
- Kill Process Tree: Terminate a selected process and its child processes.
- File Quarantine: Move a malicious file into a restricted location.
- Endpoint Isolation: Disconnect the device from other networks while retaining communication with the Hexnode XDR console.
For example, investigators examining suspicious PowerShell activity can review its parent process, command line, and associated file metadata. They can then select appropriate containment actions when malicious execution is confirmed.
XDR Containment Decision Making: Which Action Fits?
Learn when to isolate endpoints, terminate malicious processes, or quarantine suspicious files.
Hexnode UEM also provides patch compliance and vulnerability reports for enrolled devices. These include applicable vulnerability records, CVE identifiers, and patch deployment information. However, organizations must verify whether the specific ActiveMQ or Veeam updates are covered by their supported patch workflows. UEM reporting should not be presented as proof of automatic remediation for these particular server applications.
Importantly, Hexnode XDR does not establish signature detection for this specific campaign or its recovered tools. Its documented investigation and containment capabilities do not replace vendor-specific vulnerability remediation. Also, validate Windows Server eligibility separately before assuming the XDR workflow applies directly to a Qlik Sense server.
Why Should Organizations Treat Blocked Qlik Sense Attempts as Investigation Signals?
The Qlik Sense reverse shell attack demonstrates why blocked execution attempts deserve investigation beyond the initial alert.
Cynet’s findings connected suspicious Qlik Sense process activity to infrastructure containing custom backdoors, persistence scripts, and exploit-development material for ActiveMQ and Veeam. Nevertheless, the initial Qlik Sense vulnerability remains unidentified, and successful ActiveMQ or Veeam exploitation was not established in the investigated environments.
Organizations should investigate related process activity, verify persistence indicators, and remediate vulnerable ActiveMQ and Veeam installations through their respective vendor update channels. Endpoint investigation and containment can help address suspicious execution, while application-level patching addresses the underlying vulnerabilities.
Investigate and Respond to Endpoint Threats with Hexnode XDR
Explore how Hexnode XDR helps security teams investigate suspicious endpoint activity, analyze process execution, and take appropriate remediation actions.
Start Your Free Trial