As reported by The Hacker News, Belarus-aligned threat group Partisan Zmiy (Cyber Partisans) maintained persistent access inside a Russian healthcare organization for nearly two years by embedding the Vasilek backdoor within the VMware Tools installation directory — a location most EDR and endpoint monitoring tools implicitly trust and rarely inspect.
This is a textbook case of living-off-the-land tradecraft executed at a level of sophistication that should concern every defender managing virtualized infrastructure. The attackers didn't just drop a payload in a random folder; they chose a directory that endpoint agents are often configured to exclude from scanning precisely because it generates excessive telemetry noise from legitimate VMware operations. That trust assumption became the operational blind spot.
Why This Intrusion Matters Beyond the Victim
The most strategically significant detail is not the malware itself but the threat actor's restraint. Cyber Partisans have historically coupled espionage with destructive wiper attacks using their Pryanik tool. In this case, they held that capability in reserve for nearly 24 months — a deliberate decision Solar attributes to the value of the healthcare organization's trusted relationships with subsidiary medical institutions. This is trusted-relationship exploitation at scale: the compromised entity wasn't just a target, it was a launchpad.
The attackers prioritized sustained access and lateral reach over immediate impact — a hallmark of state-aligned espionage operations even when carried out under a hacktivist banner.
The technique of replacing the legitimate vmtools.dll with a malicious version containing Vasilek v1.5.8 is particularly insidious. DLL hijacking and side-loading within vendor-signed application directories bypasses application whitelisting controls that key off publisher signatures or trusted paths. The rogue authd.exe binary, launched by a masqueraded Windows service named vmauad, further demonstrates how naming conventions that mimic legitimate VMware components evade manual review by SOC analysts triaging services.
Who Is Most at Risk
Broader Implications for Defender Assumptions
The intrusion challenges a persistent assumption in endpoint hardening: that vendor application directories are inherently safe. They are not — they are simply trusted, and trust is what sophisticated adversaries exploit. The scheduled GOST tunnel operating only on Saturdays between 10–11 PM local time further illustrates beacon scheduling designed to slip past behavioral analytics that flag high-frequency or anomalous-hour C2 traffic.
Defenders should also note the systematic clearing of Windows event logs using the Evlx utility. This isn't novel, but its presence confirms the attackers anticipated forensic investigation. Log forwarding to a tamper-resistant SIEM is the only reliable countermeasure; local log stores cannot be trusted post-compromise.
Shield53 Recommendations
- Audit VMware Tools exclusions: Review all AV/EDR path exclusions. Replace blanket directory exclusions with file-hash or certificate-based allowlisting for known-good VMware binaries only.
- Implement DLL load monitoring: Deploy Sysmon Event ID 7 (image loaded) rules scoped to VMware Tools directories. Alert on unsigned or unexpectedly signed DLLs loaded by VMware processes.
- Hunt for the specific IOCs: Search for services named
vmauad, binaries namedauthd.exein VMware Tools paths, and anyvmtools.dllthat does not match VMware's known signing chain. - Detect scheduled beaconing: Tune SIEM correlation rules for narrow time-window outbound connections (e.g., weekly one-hour sessions) especially involving GOST-encrypted or Telegram-based C2 channels.
- Centralize and immutabilize logs: Forward Windows event logs to a write-once SIEM. Local log clearance via tools like Evlx should generate immediate high-severity alerts.
- Map trusted relationships: Inventory all bidirectional trust connections with subsidiaries and partners. Treat each as a potential ingress path and apply monitoring accordingly.
This intrusion is a reminder that dwell time is a defender's failure, not an attacker's achievement. Two years of undetected access in a healthcare network — where patient data and interconnected medical systems are at stake — represents a systemic monitoring gap that path-based trust models cannot close on their own.