An internet-facing firewall is supposed to stand between an attacker and the internal network. In the incident Check Point disclosed on November 12, the firewall became the attacker’s route to a VMware ESXi host. The chain began in the Mobile Access portal, gained root on the gateway, and ended with destructive activity in at least some of the investigated intrusions.
Check Point observed two waves, starting October 30 and November 2. It says a couple of dozen unpatched systems were compromised in one region in the incidents it investigated. That is a bounded observation, not an estimate of every exposed gateway. The company has not established the attacker’s identity or motive with confidence.
TL;DR
- PHP2Shell (CVE-2026-109618 and CVE-2026-109619) gave the attacker initial access through vulnerable Check Point Mobile Access portals. CVE-2026-123039 then allowed local privilege escalation through Gaia’s DynamicCLI service.
- Check Point found PHP webshells, a persistent gateway backdoor named TSGA with a TSProxy component, internal scanning, and an SSH pivot to an ESXi host where a wiper was run.
- Install the November Jumbo Hotfix on affected Security Gateways and Management Servers, prioritizing internet-facing Mobile Access. Confirm the installed hotfix and inspect the gateway and reachable internal systems for prior compromise.
- An indicator match needs investigation. A clean indicator search cannot establish that an exposed gateway was safe.
Why the gateway changed the stakes
The attacker did not need a normal administrator login to reach the Mobile Access service. Check Point’s research says PHP2Shell supplied the initial foothold, then the DynamicCLI flaw turned that foothold into root access. Root let the attacker alter the appliance itself, including files used by the portal and components that handled traffic and logs.
In the documented intrusion, the attacker placed webshells in the portal’s template directory and launched TSGA, which persisted on the gateway. Its embedded TSProxy component provided another access path through appliance processes. From the gateway, the attacker scanned internal services and reached an ESXi host over SSH using the vpxuser service account. Check Point then observed a wiper on that host. The exact path and account use are findings from this investigation; they are leads to test in your environment, not evidence that every compromised gateway led to ESXi.
What the investigation can reveal
Check Point reported PHP session files containing upload_progress_pwn, repeated Apache/PHP worker crashes, unexpected zz*.tpl files in portal templates, a process masquerading as CPconfigdaemon, and suspicious files under temporary directories. It also documented gaps in firewall logs following a logging-daemon crash. Its advisory lists two scanning IP addresses, while the research report provides file hashes and more detailed traces.
Preserve the gateway’s logs, running-process information, file metadata, configuration, and relevant network telemetry before cleanup if compromise is suspected. Establish a timeline beginning before October 30, then correlate portal requests and crashes with new template files, process starts, outbound connections, and management changes. A log gap is itself worth investigating because the observed implant could tamper with logging.
Extend the review past the gateway. Look for unusual SSH sessions from it to hypervisors and management hosts, vpxuser activity outside normal management workflows, internal scans sourced from the appliance, and newly created files or destructive commands on reached systems. Check Point documented one such path; your own segmentation and credentials determine which systems could have been reached.
What You Can Do Today
- Identify exposed gateways and their roles. List internet-facing Mobile Access portals, their Gaia versions, installed hotfix levels, Management Servers, and the internal networks each gateway can reach. Check actual external reachability rather than relying only on an inventory label.
- Apply the vendor fix. Follow Check Point’s November Jumbo Hotfix guidance for Security Gateways and Management Servers. Prioritize exposed Mobile Access deployments, use the vendor’s supported upgrade path, and verify the running appliance reports the expected fix after the change. A PHP update on an unrelated web server does not patch an affected gateway appliance.
- Reduce the reachable surface while patching. If Mobile Access is unnecessary, disable its public exposure through the supported configuration. If it is required, restrict access to approved users and networks where the deployment allows it. Limit gateway-to-management and gateway-to-hypervisor connectivity to documented operational needs, and review the credentials that permit those connections.
- Hunt for the complete chain. Search the portal and appliance for Check Point’s published indicators, then correlate with process, file, authentication and network records. Examine internal systems the gateway could access, especially ESXi and its management plane. Do not treat a missing IP or hash match as an all-clear.
- Contain a suspected breach before restoring service. Isolate the affected gateway from sensitive internal routes, preserve evidence, and involve Check Point support or incident response. Investigate credentials and keys accessible from the appliance; rotate those judged exposed through a controlled recovery plan. Rebuild or restore from a known-good image if appliance integrity cannot be established, then validate policies and logging before reconnecting it.
- Test the recovery boundary. Confirm that management access still works from approved locations, unnecessary paths to hypervisors are blocked, logs arrive continuously, and a backup or clean replacement can restore the gateway configuration. Record the time of the last known-good state for the incident review.
The hotfix closes the reported entry points. It cannot remove a webshell, a backdoor, or changes already made to a reachable host. That is why patch verification and compromise assessment belong in the same response.
Related Posts
- Check Point VPN and Management Flaws Under Attack: Patch Two Different Trust Boundaries
- Why Enterprise VPN Gateways Are Always Vulnerable
- Cookie-Controlled PHP Webshells: What Linux Hosting Logs Can Reveal
Sources
Useful read?
Find us again on Google.
Add Hive Security as a preferred source for practical security research and analysis.
Add as a preferred source on GoogleChoose Hive Security in Google's source preferences.