Microsoft's agent-framework package on PyPI was compromised on November 17th, 2026. Versions 1.26.1 and 1.26.2 contained malicious code that installs Python startup hooks via .pth files, which execute silently when the interpreter initializes. The malicious code collects credentials related to Azure, GitHub Actions, and various CI/CD environments, including environment variables, configuration files, and cloud identity tokens. It actively requests fresh Azure managed-identity and GitHub Actions OIDC tokens. The stolen data is compressed, base64-encoded, and exfiltrated to a webhook.site endpoint. The attack targets credentials from multiple tools such as pip, npm, Docker, and Git.
Multiple ELECOM wireless LAN routers (model WRC-BE36QS-B) contain an OS command injection vulnerability that allows an attacker with administrative interface access to execute arbitrary operating system commands. The affected versions are from 0 up to and including 2.1.3. The vulnerability has a high severity with a CVSS score of 7.2.
CVE-2026-130644 is a path traversal vulnerability in ansible-core affecting Red Hat Ansible Automation Platform 2 and Red Hat Enterprise Linux. The flaw occurs during collection installation with ansible-galaxy, where the install destination directory is derived from unvalidated namespace and name fields in the collection metadata. A malicious collection can exploit this to redirect installation outside the intended directory, causing deletion of writable directories and writing attacker-controlled files. Exploitation requires installing collections from untrusted sources. Red Hat plans to provide fixes for maintained streams. Mitigation includes installing collections only from trusted sources and reviewing collection metadata for suspicious path elements.
The AI Puffer – MCP Server for Claude & ChatGPT WordPress plugin is vulnerable to stored cross-site scripting (XSS) in versions up to and including 2.5.28. The vulnerability arises from insufficient sanitization of AI-generated comment replies, allowing unauthenticated attackers to inject malicious scripts via crafted comments. Exploitation requires that an administrator has enabled an auto-reply feature with replies automatically approved. This flaw can lead to script execution in users' browsers when they view the affected pages.
Gargoyle versions 2023.1 through 2026.1.1 have a stack-based buffer overflow vulnerability in the bundled Plus interpreter's ReadAction() function. This flaw occurs because the function does not properly bounds-check the index into a 1024-entry conditions array. An attacker can exploit this by supplying a crafted or truncated .plus game file with unterminated action conditions, causing out-of-bounds writes that can crash the player.
Gargoyle versions 2023.1 through 2026.1.1 have an infinite loop vulnerability in the Plus interpreter's ReadString() function. This flaw occurs because the function does not properly handle end-of-file conditions, allowing specially crafted or truncated .plus game files to cause the program to hang at full CPU usage until terminated by the user.
Pleasanter has a vulnerability in its API where sensitive information, including authentication credentials like API keys of other users and administrators, can be exposed when processing specially crafted API requests from authenticated attackers. This allows attackers to perform unauthorized actions with the privileges of compromised credentials.
fqterm versions before 0.9.11 have an out-of-bounds memory access vulnerability in the SSH client. This flaw arises because the client trusts unchecked length fields supplied by the server for memory operations. Malicious servers or man-in-the-middle attackers can exploit this by sending crafted length values during SSH protocol exchanges before host key verification, causing the client to crash.
WLA DX versions 9.7 through 10.7 have a stack-based buffer overflow vulnerability in the parse_data_blocks() function located in wlalink/analyze.c. This flaw occurs because the function copies section namespaces without proper length checks, allowing an attacker to supply a crafted object or library file with a namespace longer than 255 bytes or unterminated. This can overflow the 256-byte stack buffer during linking.
fqterm versions prior to 0.9.11 contain a heap out-of-bounds write vulnerability in the FQTermDecode function. This flaw allows malicious servers to corrupt memory by sending CSI sequences with an excessive number of semicolon-separated parameters. Specifically, attackers controlling a telnet or SSH server, or intercepting cleartext telnet traffic, can send over 30 parameters to write arbitrary 32-bit values beyond the allocated heap object.
Pleasanter has an incorrect authorization vulnerability affecting the access control of Items. This flaw allows a remote authenticated attacker with record viewing permissions to access Item values that should be restricted by Column access control settings. The vulnerability specifically impacts configurations where read access conditions are enforced via Column access control.
A stored cross-site scripting (XSS) vulnerability in GitHub Enterprise Server allowed organization members with specific permissions to inject attacker-controlled SSH keys into other members' accounts via malicious HTML in custom-property options. Exploitation required another privileged user to interact with the poisoned option and save changes with a valid sudo session. The vulnerability affected supported versions prior to 3.23 and was fixed in versions 3.19.16, 3.20.12, 3.21.10, and 3.22.5.
SOIL2 versions before 1.4.2 have a NULL pointer dereference vulnerability in their DDS, PVR, and PKM image loaders. This flaw occurs when these loaders write results through unchecked caller-supplied output pointers. An attacker can trigger a crash by supplying a valid DDS, PVR, or PKM file while passing NULL output pointers to SOIL_load_image or stbi_info.
A vulnerability in OpenShift Hive's ClusterClaim controller allows a tenant to exploit insufficient authorization checks. The controller assigns hive-claim-owner RBAC roles based on a user-controlled namespace without properly verifying the namespace against the ClusterPoolRef, enabling cross-tenant access. This flaw can let an attacker access another tenant's cluster admin kubeconfig and admin password secrets. The issue is rated medium severity with a CVSS score of 5.9. No official fix is currently available, but Red Hat recommends restricting ClusterClaim creation and updates to trusted users and validating namespaces to mitigate the risk.
A vulnerability in ASUS Router firmware allows an unauthenticated nearby attacker to perform file operations or read files outside intended directories via crafted HTTP requests during an active user session. This is due to improper authentication and improper limitation of pathname to restricted directories, commonly known as a path traversal flaw.
A SQL injection vulnerability exists in an administrative API of HCL Commerce affecting versions 8, 9.0.1, and 9.1.0 through 9.1.21. A remote attacker with administrative privileges can exploit a crafted URL to infer unauthorized data from the database. The vulnerability has a CVSS 3.1 base score of 6.5, indicating medium severity with high confidentiality impact but no integrity or availability impact.
Certain ASUS Router firmware versions have a vulnerability where sensitive information such as the router's UID or login authorization tokens can be inserted into sent data. A remote attacker can exploit this by crafting a web page that requires a logged-in router administrator to complete a third-party authorization flow, thereby exposing sensitive credentials.
Certain ASUS Router firmware versions contain a missing authentication vulnerability for a critical function. This flaw allows unauthenticated nearby users to access router information and resources, such as multimedia files, via HTTP requests. The vulnerability is identified as CWE-306 and has a medium severity rating with a CVSS 4.0 score of 5.3. No specific affected firmware versions or patch information are provided in the available data.
An XML External Entity (XXE) vulnerability exists in the Storefront REST API of HCL Commerce versions 8, 9.0.1, and 9.1.x. This vulnerability allows a remote attacker to supply specially crafted XML input that could lead to unauthorized viewing of internal files or excessive memory consumption. The vulnerability has a high severity rating with a CVSS score of 8.2.
Pleasanter has a path traversal vulnerability in its API that allows a remote authenticated attacker to access arbitrary files on the server. This occurs due to improper limitation of pathnames to restricted directories. The vulnerability has a CVSS score of 4.3, indicating medium severity.