Broken composite-type authorization checks in CockroachDB SECURITY DEFINER functions lead to unauthorized access
An attacker can exploit a flaw in how SECURITY DEFINER functions check permissions on composite parameter types. By constructing a malicious ROW object, the attacker bypasses type-level access control, allowing an unprivileged user to invoke sensitive functions with the definer's privileges and harming confidentiality and integrity.
Component
CockroachDB is a distributed SQL database designed for horizontal scalability, strong consistency, and high availability. Its architecture uses multi-tenant isolation to support large-scale transactional workloads, and its core components parse and execute complex SQL query logic.
Strengthen type-ownership validation for function arguments at the code level so that, even when a function runs as its definer, the caller's access to the types of the values passed in is still checked. Apply the principle of least privilege to block this privilege-escalation path.
CSSA Exclusive Early Warning
9.8 Critical
Parameter injection into gnuplot scripts in the OpenTSDB chart rendering component allows arbitrary command execution and remote code execution
An attacker can exploit weak validation of parameters such as key and yrange on the /q endpoint, using double URL decoding to bypass regex restrictions and inject system-call instructions containing newline characters into the gnuplot script. Because the parameters are concatenated into the script file without string escaping and the script is executed by ProcessBuilder, an unauthenticated remote attacker can execute arbitrary commands at the server operating system level.
Component
OpenTSDB is a high-performance time-series monitoring and visualization platform built on HBase. Its core architecture relies on a RESTful API to store and query massive volumes of metric data, and the built-in /q endpoint converts user-defined query conditions into gnuplot plotting instructions for multi-dimensional real-time trend analysis and charting.
Enforce strict allowlist input validation so that all user input is decoded only once and matches the expected format. Apply strict context-aware encoding or escaping to arguments passed to external programs instead of relying on blocklist filtering of special characters, and upgrade to the latest stable version that fixes this vulnerability.
CVE Intelligence
2
CVE-2026-65162
CVSS
10.0 Critical
2026-10-13
Remote code execution in NVIDIA AIStore Kubernetes PodSpec initialization
In NVIDIA AIStore for Linux, the legacy ETL Kubernetes PodSpec initialization API lacks strict validation of input parameters and contains an execution with unnecessary privileges (CWE-250) flaw. A remote attacker can exploit it to supply malicious Kubernetes host-namespace and container-security settings and bypass security restrictions. NVIDIA AIStore has a logic flaw when processing PodSpec initialization requests; an attacker can trigger it with a crafted API request, leading to remote code execution, privilege escalation, information disclosure, data tampering, and denial of service. All NVIDIA AIStore for Linux users who have not fixed this flaw are affected, including instances deployed in Kubernetes environments. An attacker only needs to send a crafted request over the network; no user interaction is required and the attack can be carried out remotely.
Component
NVIDIA AIStore is a high-performance object storage system designed for Linux, commonly used for data management in AI workloads, with Kubernetes integration.
Risks
From ordinary user to administrator: If the victim runs the browser with administrator privileges, the attacker can obtain the same privileges
Complete system control: An attacker can execute arbitrary code on the victim system and, depending on privileges, install programs, view/change/delete data, or create new fully privileged accounts
No user interaction: Through the Drive-by Compromise technique (T1189), the vulnerability can be triggered simply by visiting a malicious web page, with no further interaction
Enable sandbox isolation for the runtime environment
Block loading of Canvas scripts from untrusted sources
CVE-2026-109110
CVSS
10.0 Critical
2026-10-13
Remote command injection in Wavlink WL-WN592AX6 wmapd leads to remote code execution
In Wavlink WL-WN592AX6 firmware version 250422, the system function in the wmapd daemon component contains a security flaw involving improper neutralization of special elements used in a command (CWE-77); manipulation can lead to OS command injection. The attack can be launched remotely, and exploit code has been publicly disclosed and may be used. The vendor was contacted early in the disclosure process. All Wavlink WL-WN592AX6 devices running the affected firmware are impacted. An attacker can trigger the vulnerability remotely over the network without authentication, attack complexity is low, and the system may be fully taken over.
Component
Wavlink WL-WN592AX6 is a wireless network adapter device, and wmapd is its internal daemon component responsible for specific network mapping or management functions.
Risks
Complete system control: An attacker can execute arbitrary OS commands on the victim device and, depending on system privileges, install programs, view/change/delete data, or create new fully privileged accounts
Remote attack without interaction: An attacker only needs to send a crafted request over the network to trigger the vulnerability, with no user interaction required
Severe confidentiality and integrity impact: The CVSS score indicates severe impact on confidentiality, integrity, and availability; an attacker can steal sensitive data or disrupt normal device operation
Deploy an intrusion detection system (IDS) at the network boundary to monitor for anomalous command injection attempts against the wmapd service
Restrict network access to the device management interface and allow connections only from trusted IP addresses
Package Poisoning
8
Package Poisoning
npm2026-10-14
opencode2@>= 0
flagged as malicious
This version was found communicating with a malicious domain and executing malicious commands. Audit your dependencies and pin safe versions immediately.
MD5
3298f83035a19656eb2ce0807b6899f4
Package Poisoning
npm2026-10-14
opencode2-darwin-arm64@>= 0
flagged as malicious
This version was found communicating with a malicious domain and executing malicious commands. Audit your dependencies and pin safe versions immediately.
MD5
4add7c598e2d6059743c13046fc6a181
Package Poisoning
npm2026-10-14
opencode2-darwin-x64@>= 0
flagged as malicious
This version was found communicating with a malicious domain and executing malicious commands. Audit your dependencies and pin safe versions immediately.
MD5
e4256fb3c95e9d6f4af62ff17906bd6e
Package Poisoning
npm2026-10-14
opencode2-linux-arm64@>= 0
flagged as malicious
This version was found communicating with a malicious domain and executing malicious commands. Audit your dependencies and pin safe versions immediately.
MD5
30904307903d52edce71347ef3c3bdc6
Package Poisoning
npm2026-10-14
opencode2-linux-x64-musl@>= 0
flagged as malicious
This version was found communicating with a malicious domain and executing malicious commands. Audit your dependencies and pin safe versions immediately.
MD5
5c24a996568785e6ddbf54e5cd391802
Package Poisoning
npm2026-10-14
opencode2-windows-arm64@>= 0
flagged as malicious
This version was found communicating with a malicious domain and executing malicious commands. Audit your dependencies and pin safe versions immediately.
MD5
f0a30097bade55582dd28ab4c350c4af
Package Poisoning
npm2026-10-14
opencode2-windows-x64@>= 0
flagged as malicious
This version was found communicating with a malicious domain and executing malicious commands. Audit your dependencies and pin safe versions immediately.
MD5
523b88a13f182d30476b4ef15b8f7151
Package Poisoning
PyPI2026-10-14
sysdiagd@1.2.2
flagged as malicious
This version was found communicating with a malicious domain and executing malicious commands. Audit your dependencies and pin safe versions immediately.
