CSSA Exclusive Early Warning10.0 Critical
KiroCrew incorrect authorization allowing attackers to achieve remote code execution via forged titles
The vulnerability stems from the auto_approve_subagent_spawn predicate deciding whether to auto-approve permission events solely based on a model-generated title field, without verifying the actual event type or source identity. An attacker can craft a malicious Shell command and forge its title as a legitimate spawn_runMCP tool identifier, bypassing the hook gateway's command checks and name-grant validation. Because this logic exists in multiple transport-layer handlers and can be triggered over the network without user interaction, successful exploitation leads to unauthorized arbitrary code execution that severely damages system confidentiality, integrity and availability.
Component
KiroCrew is an automated task-processing system based on a multi-agent collaboration architecture. Its core components integrate transport channels su…
Type
Incorrect Authorization (CWE-863)
Repo
Remediation- Developers should refactor the auto-approval logic and abandon fragile checks that rely only on the title field. Introduce strict verification of canonical event identity such as tool name or non-model-generated metadata, and grant approval only when the call is confirmed as a legitimate MCP tool invocation. Also update callback contracts globally to pass complete event objects or explicit security flags to strengthen boundary protection.
CSSA Exclusive Early Warning9.6 Critical
Keycloak LDAP mapper DN parsing defect allowing unauthorized role acquisition and privilege escalation
An attacker exploits a defect in the Group-Role-LDAP-Mapper "Get Roles From User MemberOf Attribute" strategy: the system incorrectly truncates the full distinguished name (DN) in the user memberOf attribute to the first relative distinguished name (RDN) and matches by name in the global group subtree rather than validating the path. Same-named groups under different hierarchy levels are wrongly associated, allowing unauthorized users to inherit high-privilege roles and causing cross-branch lateral privilege escalation and sensitive data disclosure risk.
Component
Keycloak is a Java-based enterprise open-source identity and access management solution. Its core architecture relies on the standard OpenID Connect p…
Type
Incorrect Authorization (CWE-863)
Repo
Remediation- Fixing this vulnerability requires correcting DN handling at the code level so that LDAP queries use the full distinguished name for exact matching or post-validate that returned entries' DNs are in the user's group membership list, avoiding fuzzy name-only retrieval to ensure accurate privilege mapping.
CVE-2026-79698CVSS 9.9 Critical2026-09-07
Advantech WISE-6610 Node-RED Library command injection leading to remote code execution
In Advantech WISE-6610 series devices (including WISE-6610-NB, WISE-6610-EB, WISE-6610-TB, WISE-6610-JB, WISE-6610-CB, WISE-6610-EL-NB, WISE-6610-EL-EB, WISE-6610-EL-TB, WISE-6610-EL-JB, WISE-6610-EL-CB, WISE-6610P-DEA, WISE-6610P-DNA and WISE-6610P-DTA) version 1.2.1_20251110, the nodered_lib_apply function in the Node-RED Library component has a security defect. Improper handling of the act parameter leads to command injection (CWE-74). An attacker can launch a remote attack and execute arbitrary system commands by manipulating that parameter. Exploit code is publicly available and may be used. Upgrading to version 1.2.4_20260821 resolves the issue. Upgrading the affected component is recommended. The vendor was notified early, responded professionally, and promptly released fixed versions for the affected products.
Component
Advantech WISE-6610 series are industrial IoT gateway devices from Advantech with built-in components such as Node-RED for dataflow programming and system integration.
Risks
- Remote code execution: An attacker can execute arbitrary system commands remotely over the network without authentication and fully control the device
- High exploitability: Exploit code is public and attack complexity is low, making it easy for automated attack tools to use
- System integrity damage: An attacker can tamper with device configuration, steal sensitive data or enlist the device in a botnet
Source
Remediation- Immediately upgrade Advantech WISE-6610 series device firmware to 1.2.4_20260821 or later
- Restrict access to the Node-RED management interface at the network layer, allowing only trusted IP addresses
- Monitor device logs for anomalous system command execution behavior
CVE-2026-79697CVSS 9.9 Critical2026-09-07
Advantech WISE-6610 Basic Station Certificate-Deletion Handler command injection leading to remote code execution
In Advantech WISE-6610 series devices (including NB, EB, TB, JB, CB, EL series and P-series DEA/DNA/DTA models) version 1.2.1_20251110, the basicstation_apply function in the Basic Station Certificate-Deletion Handler component has a security defect. Improper handling of the act parameter leads to command injection (CWE-74). An attacker can launch a remote attack and execute arbitrary system commands by manipulating that parameter. The exploitation method has been publicly disclosed and there is a risk of use. Upgrading to version 1.2.4_20260821 mitigates the issue. The vendor was notified early, responded professionally, and promptly released fixed versions for the affected products.
Component
Advantech WISE-6610 series are industrial IoT gateway devices from Advantech widely used in industrial automation and edge computing for data collection, protocol conversion and edge processing.
Risks
- Remote code execution: An attacker can execute arbitrary system commands remotely without authentication and fully control the device
- High impact scope: The vulnerability affects multiple hardware models, and exploitation conditions are simple (low complexity, network-reachable), making it easy for automated attack tools to use
- Data disclosure and destruction: An attacker can read sensitive configuration, steal certificate information or disrupt normal device operation, interrupting industrial control systems
Source
Remediation- Immediately upgrade Advantech WISE-6610 series device firmware to 1.2.4_20260821 or later
- Deploy firewall rules at the network boundary to restrict unauthorized access to device management interfaces
- Monitor device logs for anomalous request behavior targeting the basicstation_apply function
CVE-2026-78080CVSS 10.0 Critical2026-09-03
JooDatabase Lite unauthenticated SQL injection leading to database compromise
In JooDatabase Lite before version 5.1.0, this Joomla extension lacks effective validation of the cid parameter and contains an unauthenticated SQL injection (CWE-89) security defect. An attacker can craft malicious requests to directly manipulate database queries, leading to sensitive data disclosure, data tampering or remote code execution. The vulnerability affects all JooDatabase Lite users who have not updated to 5.1.0, including various Joomla-based site deployment scenarios. An attacker can trigger it remotely without any authentication; the barrier is extremely low and the impact is severe.
Component
JooDatabase Lite is a database management extension for the Joomla content management system intended to simplify database operations and maintenance.
Risks
- From ordinary user to administrator: If the victim site has no additional privilege isolation, an attacker can obtain the highest database privileges via SQL injection
- Complete system control: An attacker can read, modify or delete all data in the database and, under certain configurations, achieve remote code execution on the server
- No user interaction required: An attacker need not lure users into clicking links; sending crafted HTTP requests to the target server is enough to trigger the vulnerability and automate attacks
Source
Remediation- Enforce strict input validation and filtering on critical parameters such as cid
- Deploy a web application firewall (WAF) to intercept common SQL injection attack patterns
CVE-2026-82180CVSS 10.0 Critical2026-09-03
Eclipse Arrowhead MQTT certificate authentication bypass leading to remote code execution
In Eclipse Arrowhead versions 5.0.0 through 5.2.1, when the MQTT API with a certificate authentication policy is enabled, CertificateMqttFilter has an authentication bypass by spoofing (CWE-290) security defect. The filter parses the X.509 certificate the client sends in the MQTT message payload but does not verify its signature or issuer chain, authorizing only via simple string comparison (DN-qualifier and cloud-name in the CN). An attacker can generate a self-signed certificate with crafted CN and dnQualifier values to be authenticated as the cloud system operator (isSysOp == true), bypass the downstream ManagementServiceMqttFilter and obtain full management privileges over MQTT. The vulnerability affects all unpatched Eclipse Arrowhead 5.0.0 through 5.2.1 users. An attacker only needs to publish crafted messages to the MQTT broker to complete exploitation and obtain the highest privileges without complex interaction.
Component
Eclipse Arrowhead is an open-source Internet of Things (IoT) platform designed to promote interoperability among different cloud services and devices, supporting multiple communication protocols including MQTT.
Risks
- Complete system control: An attacker can obtain system-operator privileges on the cloud platform and perform arbitrary management operations, including viewing, modifying or deleting sensitive data and controlling connected devices
- Privilege escalation: From an ordinary MQTT publisher to a system operator with full management privileges
- Remote attack: An attacker who can publish messages to the MQTT broker can trigger the vulnerability without local access or user interaction
Source
Remediation- Enforce strict access control lists (ACL) at the MQTT broker layer to limit publish permissions
- Monitor anomalous certificate authentication requests in MQTT messages, especially traffic containing self-signed certificates or specific DN patterns
CVE-2026-67276CVSS 10.0 Critical2026-09-06
RouterOS SSH RSA public-key exponent verification bypass leading to remote code execution
In RouterOS before versions 6.49.21, 7.23.4 and 7.24.2, the SSH authentication module has improper verification of cryptographic signature (CWE-347), specifically an RSA public-key exponent verification bypass. When matching an SSH authentication request to an authorized user key, the system checks only the key type and modulus and ignores the exponent. Because signature verification uses the client-supplied key, an attacker who knows an authorized RSA modulus can present a key with exponent 1, forge a valid signature and open an SSH command channel as the target user without the private key. The vulnerability affects all RouterOS users who have not updated to the fixed versions above. An attacker can launch a remote network attack and obtain SSH access to the target system without user interaction.
Component
RouterOS is a Linux-based network operating system developed by MikroTik, widely used on routers, firewalls and network management devices.
Risks
- From ordinary user to administrator: If an attacker obtains an SSH session with administrator privileges through this vulnerability, they can directly control the entire network device
- Complete system control: An attacker can execute arbitrary commands on the target device, modify network configuration, steal sensitive data or plant a backdoor
- No user interaction required: An attacker who knows an authorized RSA public-key modulus on the target device can craft malicious requests for remote exploitation without any action by the target user
Source
Remediation- Immediately upgrade RouterOS to 6.49.21 (long-term support), 7.23.4 (long-term support) or 7.24.2 (stable)
- Monitor SSH login logs to identify anomalous authentication attempts or connections from unknown sources
- Consider temporarily disabling the SSH service or restricting its access-source IPs
Package Poisoningnpm2026-09-07
omni-channel-oid-frontend@0.0.1 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
43dbf7963e03e26519dc34c2adc1f10f
Package Poisoningnpm2026-09-07
oce-configurator-wireless-frontend@9999.0.0 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
e34cb782effc6c1a06b6ba91565e48b9
Package Poisoningnpm2026-09-07
ocfe-tv-subscription-center-web@0.0.1 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
f7db766495aa8b62d32e74723faa36d5
Package Poisoningnpm2026-09-07
omni-channel-order-frontend@0.0.1 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
9bb405ced295f81a468cefb61c8d1f3f
Package Poisoningnpm2026-09-07
@caliperx2/components@>= 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
ef63fc69116971605e13b7bbc1329541
Package Poisoningnpm2026-09-07
b2b-frontend-external-library@>= 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
2ca0f468cf91acd4b5bd294654784b8b
Package Poisoningnpm2026-09-07
omni-channel-configurator-wireline-frontend@>= 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
1d61e8e326307869d4567d2663a5b3d5
Package Poisoningnpm2026-09-04
html_vue@>= 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
82898b5ab6a86ab52a3c2557182de50e
Package Poisoningnpm2026-09-04
vscode-vue@>= 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
710119459926e37c2ec325e57efcef2a
Package Poisoningpypi2026-09-04
timeweave@1.4.0 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
c63eba333b10f57f586df8a31dc386bd
Package Poisoningnpm2026-09-04
tailwind-aspect@0.4.2 flagged as malicious
This version was found communicating with a malicious domain and executing malicious commands. Audit your dependencies and pin safe versions immediately.
MD5
6965edf06cd4aedc6d821f49809a4ef3
Package Poisoningnpm2026-09-04
@crysnovax/baileys@2.7.16 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
a565c5331697d32d35015c78ff0e17b7
Package Poisoningnpm2026-09-04
megan-baileys@1.0.5 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
e08219025b36b3b1491db64a58fddb33
