Open redirect vulnerability in Discourse versions 2026.1.0 through 2026.3.0 allows unauthenticated remote attackers to redirect users to arbitrary destinations via a malicious sso_destination_url cookie, exploiting a lack of URL validation in the StaticController enter action. While the cookie is normally set during legitimate DiscourseConnect Provider flows with cryptographic validation, attackers can directly set client-controlled cookies to bypass validation logic. The vulnerability requires user interaction (clicking a crafted link) and persistence of attacker-controlled cookies to exploit, but successful exploitation can be used for credential harvesting or phishing attacks. No public exploit code or active exploitation has been confirmed at time of analysis. Patched versions 2026.1.3, 2026.2.2, and 2026.3.0 are available.
Discourse versions 2026.1.0-2026.1.2, 2026.2.0-2026.2.1, and 2026.3.0-beta allow authenticated moderators to bypass category permission controls and retrieve post content, topic titles, and usernames from categories they lack authorization to access via a sentiment analytics endpoint. Patches are available (2026.1.3, 2026.2.2, 2026.3.0); no public exploit code or active exploitation has been identified.
Server-Side Request Forgery in OpenStack Glance image import allows authenticated users to bypass URL validation via HTTP redirects and reach internal services. Affected versions include Glance prior to 29.1.1, 30.0.0 through 30.1.0, and 31.0.0. The vulnerability impacts web-download and glance-download import methods, plus the optional ovf_process plugin. An authenticated attacker can craft a redirect chain to access restricted internal endpoints, though the CVSS vector indicates no confidentiality impact and limited integrity risk (CVSS 5.0). No public exploit code or active exploitation has been confirmed at time of analysis.
Search Guard FLX versions 1.0.0 through 4.0.1 leak user credentials into audit logs when users authenticate through Kibana, exposing plaintext authentication material to any system administrator or user with log access. The vulnerability requires high-privilege access to exploit and affects only confidentiality, but the presence of credentials in audit logs creates a persistent information disclosure risk that persists across backup and archival systems.
Stored Cross-Site Scripting (XSS) in SonicWall Email Security allows authenticated admin users to inject and execute arbitrary JavaScript code through improper input sanitization during web page generation. The vulnerability affects all versions of SonicWall Email Security appliance and requires admin-level authentication to exploit, limiting immediate exposure but posing significant risk to organizations where admin accounts are compromised or insider threats exist.
Stored cross-site scripting (XSS) in Pega Platform versions 8.1.0 through 25.1.0 allows authenticated administrative users with extensive access rights to inject malicious scripts into user interface components, potentially compromising the confidentiality of other users who interact with affected UI elements. The vulnerability requires high-privilege administrative access and user interaction to exploit, resulting in a CVSS 4.8 (low severity) with no integrity or availability impact. No public exploit code or active exploitation has been identified at time of analysis.
HTTP Request Smuggling in cpp-httplib prior to 0.40.0 allows remote attackers to inject arbitrary HTTP requests on HTTP/1.1 keep-alive connections by embedding malicious request data in the body of GET requests that the static file handler does not consume. The unread body bytes remain on the TCP stream and are interpreted as a new request, enabling information disclosure and request manipulation without authentication or user interaction.
Anthropic Python SDK versions 0.86.0 to before 0.87.0 create memory files with overly permissive file permissions (0o666), allowing local attackers to read persisted agent state or modify memory files to influence model behavior on shared hosts and Docker environments. The vulnerability affects both synchronous and asynchronous memory tool implementations and has been patched in version 0.87.0; no public exploit code or active exploitation has been identified at the time of analysis.
DNSdist instances using custom Lua code can be crashed via denial of service when the DNSQuestion:getEDNSOptions method accesses a modified DNS packet, triggering a use-after-free condition. This affects DNSdist across all versions and requires network access to send crafted DNS queries, but the attack demands specific Lua code patterns and high attack complexity; no public exploit or active exploitation has been confirmed, and the real-world impact is limited to environments where custom Lua DNS query handlers reference EDNS options.
DLL search-order hijacking in Anthropic Claude for Windows installer (Claude Setup.exe) versions before 1.1.3363 enables local privilege escalation to system context. An attacker with low privileges and physical or local access can plant a malicious DLL (such as profapi.dll) in the installer directory; when an elevated user runs the installer, the uncontrolled search path causes the malicious DLL to be loaded and executed with system privileges, achieving arbitrary code execution. No public exploit code or active exploitation has been confirmed at the time of analysis.
Admidio 5.0.0 through 5.0.7 allows authenticated users to permanently delete list configurations via CSRF attacks in the mylist_function.php delete handler, lacking CSRF token validation. An attacker can craft a malicious page to silently destroy a victim's shared list configurations, including organization-wide lists if the victim holds administrator rights. No public exploit code has been identified at time of analysis. Vendor-released patch: version 5.0.8.
Admidio prior to version 5.0.8 allows attackers with pending registration status to bypass CSRF protections and trick administrators with approval rights into automatically approving registrations via malicious URLs, enabling unauthorized account activation without manual review. The vulnerability affects the create_user, assign_member, and assign_user action modes in modules/registration.php, which process GET requests without token validation unlike the delete_user mode in the same file. An attacker extracts their user UUID from a registration confirmation email, crafts a URL targeting administrators, and gains illicit account approval through social engineering rather than technical compromise.
Discourse versions 2026.1.0 through 2026.1.2, 2026.2.0 through 2026.2.1, and 2026.3.0 prior to patching allow authenticated users to disclose shared draft topic titles via specially crafted inline onebox requests that reference the shared drafts category. An attacker with valid Discourse credentials can enumerate and read draft titles not intended for their access, violating information confidentiality. The vulnerability has been patched in versions 2026.1.3, 2026.2.2, and 2026.3.0; EPSS and KEV data indicate no current active exploitation, though the fix is available and should be deployed promptly given the low barrier to exploitation.
Discourse chat user search functionality discloses channel membership information to authenticated users without proper authorization checks, allowing users to infer private channel membership across versions 2026.1.0-2026.1.2, 2026.2.0-2026.2.1, and 2026.3.0-rc1, affecting community administrators and organizations relying on channel privacy. The vulnerability requires authenticated access but carries low confidentiality impact (CVSS 4.3); patches are available in versions 2026.1.3, 2026.2.2, and 2026.3.0.
Open redirect in Search Guard FLX up to version 4.0.1 allows unauthenticated remote attackers to craft malicious requests that redirect users to untrusted URLs, enabling phishing and credential theft attacks. The vulnerability requires user interaction (clicking a redirected link) and affects all versions through 4.0.1. No public exploit code or active exploitation has been confirmed at time of analysis.
User Profile Builder plugin for WordPress up to version 3.15.5 allows authenticated subscribers and above to reassign ownership of arbitrary posts and attachments through insecure direct object reference (IDOR) in the wppb_save_avatar_value() function. The vulnerability lacks validation on user-controlled keys, enabling privilege escalation where low-privileged users can modify post_author fields to take control of content created by other users. No public exploit code or active exploitation has been identified at time of analysis.
WWBN AVideo versions 26.0 and prior allow authenticated uploaders to bypass content moderation by directly setting video status to active via an unvalidated overrideStatus parameter, circumventing admin-controlled review workflows. The vulnerability affects any user with upload permissions and has a CVSS score of 4.3 (low-to-moderate severity) with no public exploit code identified at time of analysis.
Admidio versions prior to 5.0.8 allow authenticated users to bypass CSRF token validation and server-side form validation in the inventory module's item_save endpoint by setting the imported POST parameter to true, enabling unauthorized modification of inventory item data without proper security checks. The vulnerability requires valid authentication but carries moderate impact due to the complete circumvention of two independent security controls.
Local integrity modification in iccDEV prior to version 2.3.1.6 affects the CIccCLUT::Iterate() function and CLUT dumping output in CIccMBB::Describe(), allowing local attackers without privileges to alter ICC color profile data integrity. The vulnerability requires local access and produces incorrect LUT (Look-Up Table) dump output that could compromise color management workflows relying on accurate profile representation.