Critical vulnerabilities, curated daily for security professionals
Friday's disclosures center on widely deployed infrastructure and web platforms, including a maximum-severity flaw in MariaDB Server and high-impact issues across Ubiquiti's UniFi OS and UID Enterprise Agent, cPanel's WordPress Toolkit, and the ClipBucket video platform. The brief covers 12 critical CVEs, unchanged from the prior day, alongside 61 high-priority vulnerabilities, a 65% increase. Notable entries include CVE-2026-49261 (CVSS 10, MariaDB Server), CVE-2026-47369 (CVSS 9.9, Ubiquiti UniFi OS), and CVE-2026-47365 (CVSS 9.9, cPanel WordPress Toolkit). Remote code execution, authentication bypass, and unauthenticated access dominate the affected products, spanning database, network management, hosting, and WordPress ecosystems. No patches were available at disclosure time for the tracked critical issues, so teams should prioritize compensating controls and monitor vendor advisories closely; seven CVEs carry confirmed active exploitation.
Immediate action: Prioritize MariaDB Server, Ubiquiti UniFi OS and UID Enterprise Agent, cPanel WordPress Toolkit, and ClipBucket deployments for immediate review, and apply compensating controls for actively exploited products including Ivanti Sentry, Check Point Security Gateway, SolarWinds Serv-U, and Cisco Catalyst SD-WAN Manager. With no patches available for the tracked critical issues at disclosure, restrict network exposure of affected services and monitor vendor channels for fixes.
CVSS score (e.g. 9.1) — severity from 0–10. Red marks critical (9+), orange high (7–8.9).
Exploitability — how hard the flaw is to attack, read from the CVSS vector:
The lower the bar on all three, the easier to exploit at scale — “Network · No privileges · No interaction” is the worst case: hit from anywhere, no credentials, no victim action.
🔴 Actively exploited — confirmed under attack in the wild (CISA’s Known Exploited Vulnerabilities catalog). Prioritize these regardless of score.
EPSS · Nth percentile — FIRST.org’s estimated chance a flaw is exploited within 30 days. We flag it only in the top 10% — a statistical signal it’s unusually likely to be targeted, separate from whether attacks are confirmed.
Check Point Security Gateway is affected by an improper authentication vulnerability that is currently being exploited in the wild.
A critical OS command injection vulnerability in Ivanti Sentry allows remote unauthenticated users to achieve root-level remote code execution.
SolarWinds Serv-U is vulnerable to an uncontrolled resource consumption flaw allowing unauthenticated attackers to crash the service via specially crafted POST requests.
LiteLLM contains a command injection vulnerability in its MCP server test endpoints that, when chained with a host header bypass, enables unauthenticated remote code execution.
An out-of-bounds read and write vulnerability in the V8 JavaScript engine allows remote attackers to execute arbitrary code via a crafted HTML page.
Arista Extensible Operating System is affected by an incomplete comparison vulnerability, currently tracked in the CISA KEV catalog.
A command injection and privilege escalation vulnerability exists in the CLI of Cisco Catalyst SD-WAN Manager due to insufficient input validation.
An argument injection vulnerability in WordPress Toolkit allows authenticated users to bypass cross-tenant authorization and execute arbitrary CLI commands as another account.
The ClipBucket video sharing platform is vulnerable to unauthenticated blind SQL injection, allowing attackers to execute unauthorized database queries.
A SQL injection vulnerability in thaipalliative_lte allows remote attackers to execute arbitrary SQL commands via unsanitized input in the idFormMain and id parameters.
A command injection vulnerability in MariaDB server allows shell command execution via the joiner node name when the `wsrep_notify_cmd` configuration is enabled.
An incorrect privilege assignment vulnerability in Hippoo Mobile App for WooCommerce allows remote attackers to perform privilege escalation.
An unrestricted file upload vulnerability in Başarsoft Rotaban allows remote attackers to upload and execute a web shell on the server.
Limatek System LimRAD NAC contains an unrestricted file upload vulnerability that allows for remote code inclusion and arbitrary code execution.
ClipBucket v5 contains a command injection vulnerability in the Remote Play feature, allowing authenticated users to execute arbitrary shell commands via unsanitized URL inputs.
The Ubiquiti UID Enterprise Agent contains an improper input validation vulnerability that allows low-privileged network attackers to execute arbitrary commands on the host device.
Certain Ubiquiti devices running UniFi OS are susceptible to an improper input validation vulnerability that allows network-based attackers to escalate privileges on the device.
Certain Ubiquiti devices running UniFi OS contain an improper input validation vulnerability allowing low-privileged network attackers to execute arbitrary commands.
Improper authentication checks in the OAuth implementation of the affected software allow for account hijacking, even when the feature is disabled.
A use-after-free vulnerability in the MongoDB Server JavaScript engine allows authenticated users to trigger memory corruption.
A use-after-free vulnerability in the core of Google Chrome on Windows allows for potential arbitrary code execution.
A use-after-free vulnerability exists in the Media component of Google Chrome on Windows, potentially allowing remote code execution or system compromise.
A use-after-free vulnerability in the Autofill component of Google Chrome on Mac could allow a remote attacker to trigger heap corruption via a crafted HTML page.
A use-after-free vulnerability in DigitalCredentials in Google Chrome allows a compromised renderer process to potentially perform a sandbox escape.
A heap buffer overflow in the GPU component of Google Chrome on Android allows a compromised renderer process to potentially perform a sandbox escape.
A use-after-free vulnerability in the WebMIDI component of Google Chrome on Windows allows an attacker to potentially escape the sandbox.
A use-after-free vulnerability exists in the Cast component of Google Chrome, potentially allowing for a sandbox escape via malicious network traffic.
A heap buffer overflow in the Codecs component of Google Chrome for Linux and ChromeOS could allow a remote attacker to perform a sandbox escape.
A use-after-free vulnerability in the GPU component of Google Chrome for Mac allows for potential sandbox escape via crafted HTML content.
A use-after-free vulnerability in the GPU component of Google Chrome for Android could allow a remote attacker to perform a sandbox escape.
A use-after-free vulnerability in the Video component of Google Chrome for Windows allows for potential sandbox escape via crafted HTML content.
A heap buffer overflow vulnerability in the GPU process of Google Chrome on Android may allow for sandbox escape and privilege escalation.
Apache OFBiz contains a template injection vulnerability that allows authenticated users with specific privileges to execute arbitrary code.
An inappropriate implementation in the Mojo inter-process communication system within Google Chrome on Windows may lead to security boundary violations.
A type confusion vulnerability in Microsoft Office allows an unauthorized attacker to execute arbitrary code locally on an affected system.
A type confusion vulnerability in Microsoft Office allows an unauthorized attacker to execute arbitrary code locally on an affected system.
A type confusion vulnerability in Microsoft Office allows an unauthorized attacker to execute arbitrary code locally on an affected system.
Insufficient validation of untrusted input in the Accessibility feature of Google Chrome on Mac allows for potential sandbox escapes.
An inappropriate implementation in Google Chrome's DevTools component allows for a sandbox escape via a crafted HTML page.
A race condition in the Safe Browsing feature of Google Chrome on Mac allows for a sandbox escape via a malicious file.
An out-of-bounds write vulnerability in the GPU component of Google Chrome on Android allows for potential sandbox escapes.
An inappropriate implementation in the Views component of Google Chrome on Windows allows for a sandbox escape.
Google Chrome on Linux contains a vulnerability due to insufficient validation of untrusted input within the Linux Toolkit Theming component.
Insufficient validation of untrusted input in the Dawn graphics component of Google Chrome on Linux and ChromeOS allows for potential sandbox escapes.
An inappropriate implementation in the Views component of Google Chrome on Linux allows for potential sandbox escapes.
An out-of-bounds read vulnerability in Microsoft Office Excel allows an unauthorized attacker to disclose sensitive information over a network.
A vulnerability exists in the Roxy-WI interface, which is used for managing Haproxy, Nginx, Apache, and Keepalived servers.
A vulnerability in LiteLLM prior to version 1 allows for potential unauthorized actions or information disclosure within the platform.
Axios is vulnerable to a Prototype Pollution attack that can be escalated into a full Man-in-the-Middle (MITM) attack.
Axios fails to properly validate IPv4-mapped IPv6 addresses in its proxy bypass function, leading to potential SSRF.
Improper authentication in Windows Cryptographic Services allows an unauthorized attacker to perform local privilege escalation.
An out-of-bounds read vulnerability in Windows Hyper-V allows an authenticated attacker on a guest VM to execute arbitrary code on the host server.
A type confusion vulnerability in Windows Hyper-V allows an authenticated attacker on a guest VM to execute arbitrary code on the host server.
An access control bypass in mcp-server-kubernetes allows clients to invoke Kubernetes operations directly, bypassing tool discovery restrictions.
A critical vulnerability has been identified in the ClipBucket open-source video sharing platform.
A security vulnerability in Fission allows unauthorized modification of container capabilities, potentially leading to node-level clock corruption.
An improper access control vulnerability in the CelloOS SSH service allows authenticated remote attackers to bypass command restrictions and execute unauthorized OS commands.
Fission contains a namespace validation flaw in its admission webhook, allowing users to bypass security boundaries when deploying functions.
A vulnerability in OpenClaw's interactive callback functionality poses a significant security risk to users.
A stored cross-site scripting (XSS) vulnerability in the Analytics Dashboard allows authenticated developers to execute malicious code in the context of other users.
An improper authorization flaw in Group SAML identity management allows authenticated group owners to perform account takeovers of other group members.
A privilege escalation vulnerability in IBM i allows unauthorized users to execute code with administrator privileges via an unqualified library call.
A protection mechanism failure in macOS Sequoia 15.4 allows an application to break out of its sandbox, potentially leading to unauthorized system actions.
OpenClaw contains a shell option parsing vulnerability that allows combined POSIX shell flags to bypass exec revalidation checks.
A code execution vulnerability in OpenClaw allows marketplace runtime extension metadata to redirect loading toward unscanned package payloads.
OpenClaw contains a privilege escalation vulnerability in the Matrix allowFrom feature that allows authenticated accounts to match policy entries through mutable display name metadata.
A locality validation vulnerability in the OpenClaw Control UI pairing allows network-adjacent attackers to obtain durable admin-capable device tokens.
An arbitrary code execution vulnerability exists in OpenClaw skill install flows where workspace .env files can override the Homebrew executable selection.
LiteLLM contains a privilege escalation vulnerability allowing authenticated users to generate unauthorized API keys, granting them full proxy administrative access.
A path traversal vulnerability in UniFi OS allows an attacker with network access to read sensitive data from the device.
Slate Digital Connect for macOS contains an improper certificate chain of trust validation flaw in its XPC Service, enabling local privilege escalation via malicious client signing.
Slate Digital Connect for macOS contains a TOCTOU race condition in its PID-based client validation, allowing local attackers to escalate privileges.
Adobe ColdFusion contains an incorrect authorization vulnerability that allows a high-privileged attacker to achieve arbitrary code execution.
Adobe ColdFusion contains an improper input validation vulnerability that allows high-privileged attackers on an adjacent network to execute arbitrary code.
A cross-workspace asset authorization bypass exists in Plane, allowing authenticated users to read, copy, delete, and overwrite assets in other workspaces.
OpenClaw versions prior to 2026 contain a critical security vulnerability requiring immediate remediation.
The Metrics::Any::Adapter::Statsd library contains a vulnerability affecting versions prior to the current release.
An initialization error in Wss4jSecurityInterceptor disables WS-I Basic Security Profile enforcement, potentially causing services to accept invalid security messages.
The Jaxp13XPathTemplate component uses an insecure, unhardened XML parser for XPath expressions, potentially exposing applications to XML External Entity (XXE) attacks.
A type-confusion vulnerability in node-tmp allows attackers to perform path traversal by supplying non-string values to file path parameters.