Apache Airflow's scheduler deserializes XCom values written by task code when evaluating task dependencies. This behavior allows a Dag author, who controls XCom data, to influence the scheduler's deserialization process, which is intended to be restricted to code installed by the Deployment Manager. This can lead to privilege escalation in deployments where Dag authors are less trusted than Deployment Managers. The issue is addressed in Apache Airflow version 3.4.0 and later by restricting dependency evaluation to primitive XCom values.
A stored cross-site scripting (XSS) vulnerability exists in the WP Customer Area plugin by Aguila Technologies, affecting versions up to 8.4.3. This vulnerability arises from improper neutralization of input during web page generation, allowing malicious scripts to be stored and executed in users' browsers.
An OAuth 1 rebinding vulnerability exists in Apache Allura versions through 1.22.0. This security flaw has been addressed in version 1.22.1, which users are advised to upgrade to in order to mitigate the issue.
Apache Airflow versions before 3.3.0 are not affected by CVE-2026-96285. Version 3.3.1 includes a partial fix that checks the scheduler's decoded next_kwargs value is a dictionary only after deserializing the entire stored payload. This deserialization process reconstructs nested objects and runs their constructors before the check, allowing a DAG author to trigger constructor execution in the scheduler and API server without non-default configuration. The fix in 3.3.1 is incomplete; deployments must upgrade to version 3.4.0 or later for a complete resolution.
A missing authorization vulnerability exists in Mamunur Rashid Radius Booking — Booking Calendar for Appointments & Services up to version 1.0.20. This flaw allows attackers to bypass access control mechanisms due to incorrectly configured security levels.
The FraudLabs Pro SMS Verification WordPress plugin versions 1.11.2 through 1.12.1 have a vulnerability where they do not perform authorization or validity checks before marking a WooCommerce order's SMS verification as completed. This flaw allows unauthenticated users to forge the verified state of an order, effectively bypassing the SMS verification control.
A security vulnerability exists in iomad versions up to 4.5.15, 5.0.11, 5.1.8, and 5.2.4 affecting the local_iomad_pluginfile function in the PDF Certificate Handler component. This flaw allows remote attackers to bypass authorization controls. A patch has been identified to address this issue. The vulnerability has a medium severity rating with a CVSS score of 4.3.
The Confection WordPress Cable plugin versions 0.4.0 through 0.4.7 lack capability and nonce checks when saving settings, allowing unauthenticated users to write arbitrary PHP code to a file that the plugin serves. This enables remote code execution on the server without authentication.
The JWT Auth WordPress plugin versions through 3.0.2 have a vulnerability where previously issued refresh tokens are not invalidated when a user changes or resets their password or removes a connected device. This allows an attacker who obtained a refresh token before such an event to continue issuing valid access tokens and maintain authenticated access to the affected account, including accounts with Administrator privileges.
Apache Airflow's CLI audit logging masks secrets based on argument position, but some command argument patterns leave secrets exposed in cleartext within audit logs. This affects commands such as Variable commands with trailing --description, --conn-extra, and inline --password in airflow keycloak-auth-manager commands. Users with read access to audit logs but without Variable or Connection permissions can retrieve sensitive secrets via the event-log API. Users are advised to upgrade to version 3.4.0 or later and to purge old audit log entries containing exposed secrets and rotate those secrets.
Apache Airflow's public Import Errors API exposes raw parse-failure stack traces to any user with the ImportError.can_read permission when the failing file has no registered DAG. This can reveal sensitive information such as file paths and code to unauthorized users. The issue is addressed in Apache Airflow 3.4.0 and later by restricting stack trace visibility to users with the new IMPORT_ERRORS_ALL permission.
A stored cross-site scripting (XSS) vulnerability exists in StylemixThemes Motors plugin up to version 1.4.128. This vulnerability allows improper neutralization of input during web page generation, potentially enabling an attacker with low privileges to execute malicious scripts in the context of other users. The CVSS score is 6.5, indicating a medium severity impact with partial confidentiality, integrity, and availability loss.
Nimbus in Apache Storm instantiates a class specified by a topology's scheduler strategy configuration. The intended allowlist to restrict which classes can be instantiated was unset by default, effectively allowing any class on the Nimbus classpath with a no-argument constructor to be instantiated. This allows an authenticated submitter to cause Nimbus to instantiate classes with side effects, such as starting additional ZooKeeper servers or local clusters, leading to resource exhaustion in the Nimbus daemon. The issue is fixed in Apache Storm 3.1.0 by changing the default behavior to only allow built-in strategies when the allowlist is unset. Clusters using out-of-tree scheduler strategies must explicitly whitelist those classes or face rejection of the topologies.
Apache Airflow versions with multi-team deployments and secret cache enabled have a vulnerability where secret cache keys are constructed using unescaped underscore delimiters from team names and secret IDs. This can cause key collisions allowing a DAG author to access another team's Variable or Connection values. The issue affects deployments with both core.multi_team = True and secrets.use_cache = True. Upgrading to Apache Airflow 3.4.0 or later or disabling the secret cache mitigates the issue.
CVE-2026-124587 is a medium severity information disclosure vulnerability in elizaOS eliza 2.0.3-beta.7 affecting an unknown function in the HTTP Server component. The flaw allows remote attackers to obtain sensitive information. Public exploit code is available. A patch identified by commit cf6fa193017e70884e01c5bb7890f74b6b2b79d2 exists and should be applied to remediate the issue.
A code injection vulnerability exists in huggingface finetrainers versions 0.1.0 and 0.2.0 within the resolve_component_cls function of finetrainers/utils/model.py. This flaw allows remote attackers to manipulate the Library argument to execute arbitrary code. The vulnerability has a CVSS score of 6.3 (medium severity) and public exploit code is available. The project has been notified but has not issued a response or patch yet.
Apache Airflow's audit logs stored secret values from Connections and Variables in cleartext within the Log.extra column. This occurred because the secrets masker only matched configured sensitive key names, failing to mask all secret values. An authenticated user with low privileges and read access to the audit API could retrieve these secrets without prior access to the underlying resources. The issue is fixed in Apache Airflow version 3.4.0, which masks secret values properly and restricts audit log access to authorized users.
Tencent RapidJSON version 1.1.0 contains a vulnerability in the GetCachedPowerByIndex function within include/rapidjson/internal/diyfp.h. The flaw involves manipulation of the decimalPosition argument, leading to an out-of-bounds read. This vulnerability can be exploited remotely without authentication. Public exploit code has been disclosed. The project has been notified but has not yet issued a response or patch.
A reflected Cross-site Scripting (XSS) vulnerability exists in the Review Schema software up to version 3.1.4. This vulnerability allows improper neutralization of input during web page generation, potentially enabling an attacker to execute arbitrary scripts in the context of a user's browser session.
The DefaultStateSerializer component in Apache Storm, used for persisting and restoring stateful bolt checkpoints with a Redis-backed state provider, improperly configures Kryo serialization to allow deserialization of untrusted data without class registration. This enables an attacker with write access to the state store keyspace to inject serialized objects that execute arbitrary code in the worker JVM during state restoration. The vulnerability requires Redis-backed state usage and write privileges to the state store, which is a data-plane privilege distinct from code execution rights. Apache Storm 3.1.0 addresses this by enforcing class registration in the serializer, but upgrading may require migration due to changes in checkpoint restoration behavior.