Skip to main content
Mallory
HighExploited in the wildCISA KEVPublic exploitVulnerability

CVE-2026-123039: Local Privilege Escalation in Check Point Gaia OS DynamicCLI

Published Nov 12, 2026 · Updated today
Trending this week#4of 427,564 vulnerabilities
Mention activity · 30 days157 mentions · peak 64 in a dayToday
CVSS7.8High · v3.1
EPSS2.2%83rd percentile
CISA KEVYesAdded Nov 13, 2026
In the wildActivePer CISA KEV
IdentifiersCVE-2026-123039CWE-732 · Incorrect Permission Assignment…

CVE-2026-123039 is a local privilege-escalation vulnerability in the DynamicCLI service of Check Point Gaia OS. The service runs as root and accepts commands through a local UNIX-domain socket created with world-writable permissions, without authenticating or authorizing connecting processes. A low-privileged local user or compromised service account can submit commands that execute as root. Affected products include Check Point Security Gateways, management and logging systems, and Maestro Orchestrator. Affected releases include R81, R81.10, R81.20, R82, R82.10, and R82.20; Spark Firewalls and Smart-1 Cloud are unaffected. The vulnerability has a CVSS v3.1 score of 7.8 and is actively exploited, including in chains with the PHP2Shell vulnerabilities CVE-2026-109618 and CVE-2026-109619.

Do you run this?See affected versions in your environment and whether adversaries are exploiting this vulnerability against assets like yours.
Check your exposure
Analyst brief

Impact, mitigation & remediation

What it means and what to do now: patch path, interim mitigations, and the assume-compromise checklist.

Impact

What an attacker gets, and what they’ve been doing with it.

Successful exploitation compromises the confidentiality, integrity, and availability of the affected appliance by granting root-level command execution. Attackers can access sensitive configurations, modify security policies, intercept traffic, install persistent implants, tamper with logs, or disrupt gateway operation. Observed attack chains deployed TSGA and TSProxy and, in some incidents, subsequently accessed VMware ESXi hosts and destroyed data. Those downstream outcomes require additional access and attacker actions; they are not direct effects of this local vulnerability alone.

Mitigation

If you can’t patch tonight, do this now.

If immediate patching is not possible, use Check Point's documented workaround to stop dynamic_server and prevent its restart. This disables Dynamic CLI commands and is temporary, not a replacement for patching. Restrict administrative access to trusted systems and minimize accounts with local shell access. Review DynamicCLI activity for unexpected commands, account changes, downloads, and access outside the expected Clish workflow. Investigate unexplained logging gaps and outbound SSH connections to internal hypervisors; rotate potentially exposed administrative credentials when compromise is suspected.

Remediation

Patch, then assume compromise.

Install the November Jumbo Hotfix or a later applicable Jumbo Hotfix Accumulator. Minimum fixed levels are R81.20 Take 174, R82 Take 132, R82.10 Take 52, and R82.20 Take 14. Move unsupported R81 and R81.10 deployments to supported, fixed releases. The correction restricts the DynamicCLI socket to permissions 0660 with root:config ownership and blocks and logs unauthorized requests. Apply the PHP2Shell fixes as well: the standalone October PHP2Shell hotfix does not fix CVE-2026-123039. Investigate previously exposed appliances and connected systems because patching does not remove existing implants or invalidate stolen credentials.

Public exploits

Exploits

1 exploit validated by Mallory's analysis. Hidden: 1 rejected by Mallory's analysis as not a working exploit.

View more in app →
CVE-2026-123039This standalone repository contains a working local privilege-escalation demonstration against an explicitly synthetic model of Check Point Gaia OS Dynamic CLI. It attributes the vulnerability to CVE-2026-123039 and CWE-732/CWE-862: a root daemon exposes a world-writable UNIX socket without caller authorization, allowing a low-privilege local process to submit commands for execution as root. The supplied material does not independently establish the CVE record, vendor advisory, affected-version claims, or compatibility with actual Gaia binaries. The 63-file tree contains 13 source/build files, counting the C header, Makefile, and two Containerfiles. poc/poc.py is the standard-library Python entry point with check, exploit, probe, and mitigation modes. It supports configurable socket and marker paths, timeouts, JSON output, and a root-execution guard. Even check mode actively executes id -u rather than relying only on permissions. The provided exploit payload is a basic root-owned marker demonstration, supporting OPERATIONAL maturity rather than a customizable exploitation framework. lab/src/ contains the C daemon model, shared request parser, deterministic sanitizer harness, and Bash configuration, supervision, startup, and sweep utilities. Four authorization variants separate the effects of filesystem permissions and peer authorization: vulnerable (0777/no validation), fixed (0600/root-only peer validation), permissions-only, and peer-validation-only. run-lab.sh and Makefile orchestrate Podman builds, shared-volume execution, differential tests, mitigation experiments, and evidence collection. The container network is laboratory infrastructure; exploitation itself uses AF_UNIX, not TCP/IP. EXPOSE 22 is image metadata and does not establish an SSH listener or remote exploit path. Supplied captures report uid-1000 callers obtaining uid-0 command output and a root-owned marker on the vulnerable model. The fixed and permissions-only models refuse connections with EACCES; the peer-only model accepts a connection but denies execution. Captures also show that disabling the modeled daemon withdraws the socket, whereas manually changing socket permissions is undone on restart. These results are repository-provided evidence, not independently executed tests. The parser accepts framed and legacy commands, rejects empty, oversized, truncated, and embedded-NUL commands, and ignores bytes after the declared payload. A notable reporting limitation is that the shorter-length probe is labeled blocked despite returning root identity output, so its classification can miss successful execution of a truncated command. The reported 200012 clean sanitizer invocations apply only to the model parser and tested corpus, not to the proprietary product or the complete daemon. The supervisor also reads the live configuration file directly, so the model does not strictly require :save before noticing an on/off change. No reverse shell, credential theft, persistence, or malicious external callback is visible. GitHub requests are confined to publication metadata checks. Several important source sections are truncated, including the marker-generating command and peer-validation handler, limiting complete verification. Documentation contains a stale 37-file publication claim, while the supplied inventory lists 63 files. The archive size was not provided; size_bytes is 0 to represent unavailable metadata.googlesky · Disclosed Nov 13, 2026 · python · c · localOperational
FAQ

Frequently asked questions

Is CVE-2026-123039 exploited in the wild?
Yes. As of Nov 16, 2026, Mallory has reports of exploitation in the wild, and CISA added it to KEV on Nov 13, 2026.
Is there a public exploit for CVE-2026-123039?
Yes. Mallory tracks 1 public exploit or proof of concept for CVE-2026-123039, the earliest disclosed Nov 13, 2026.
Is there a patch for CVE-2026-123039?
Install the November Jumbo Hotfix or a later applicable Jumbo Hotfix Accumulator. Minimum fixed levels are R81.20 Take 174, R82 Take 132, R82.10 Take 52, and R82.20 Take 14.
How severe is CVE-2026-123039?
CVE-2026-123039 has a CVSS v3.1 score of 7.8 (High) and an EPSS score of 2.2% (83rd percentile) as of Nov 16, 2026.
Activity feed

Recent activity

158 sources tracked across advisories, community write-ups, and news. New activity surfaces here as Mallory finds it.

View more in app →
cyberveilleNews · Nov 16, 2026🪲 Semaine 46 - CVE les plus discutées | CyberVeilleThe Gaia OS DynamicCLI daemon accepts commands through an unauthenticated, world-writable local socket and executes them as root, enabling authenticated local users to escalate privileges. The page identifies Quantum Security Gateway as affected, reports CISA KEV inclusion and proof-of-concept material, and recommends restricting socket access.Read more ↗malware newsNews · Nov 16, 202616th November - Threat Intelligence Report - Malware News - Malware Analysis, News and IndicatorsA privilege-escalation vulnerability exploited after initial access through PHP2Shell on Check Point Security Gateways. The campaign initially targeted Israeli organizations before expanding worldwide.Read more ↗checkpoint research blogNews · Nov 16, 202616th November - Threat Intelligence Report - Check Point ResearchA privilege-escalation vulnerability exploited after attackers gained access through PHP2Shell on Check Point Security Gateways. The campaign achieved persistence and sometimes wiped a VMware ESXi host.Read more ↗security weekNews · Nov 16, 2026Recent UnrealIRCd Vulnerability Exploited in Attacks - SecurityWeekA high-severity vulnerability affecting multiple Check Point products that has been exploited in destructive attacks against Israeli organizations. Mentioned briefly as another vulnerability added to CISA's Known Exploited Vulnerabilities catalog; technical details and remediation availability are not provided.Read more ↗
+30 more in the timeline. Sign up to see them
Do you run this?See affected versions in your environment and whether adversaries are exploiting this vulnerability against assets like yours.
Check your exposure
Mallory platform

The version that knows your environment.

This page is what’s public. Mallory adds the parts that aren’t: which of your assets are affected, which adversaries are exploiting it right now, and which detections to deploy tonight.
Exposure mappingQuery your assets running an affected version, and investigate the blast radius.
Threat actor evidenceEvery report linking this CVE to a named adversary.
Associated malware49Malware families linked to this CVE, with evidence and IOCs.
Detection signatures1YARA, Sigma, Snort, and vendor rules, auto-deployed to your SIEM.
Vendor-by-vendor mappingCross-references every affected SKU, including bundled OEM variants.
Social activity121Community discussion across Reddit, Mastodon, and other social sources.