Vikunja versions prior to 2.7.1 contain a vulnerability that allows authenticated users to exhaust server memory by uploading multiple concurrent images. The flaw arises from uncontrolled resource consumption during the decoding of highly compressible PNG images, which can reach up to 50 megapixels each. This can cause the API process to experience memory pressure or be terminated by the operating system due to out-of-memory conditions.
Vikunja versions before 2.7.1 have a vulnerability where dump archives created by the vikunja dump command have insecure default file permissions. This allows local users on a shared host to read these world-readable zip files, potentially exposing sensitive data such as configuration files, password hashes, TOTP secrets, tokens, and stored files.
Vikunja versions before 2.7.1 have a credential exposure vulnerability where malicious repository authors can specify an attacker-controlled server in the .veans.yml file. When veans runs in an untrusted checkout, it may send sensitive credentials such as VEANS_TOKEN, login passwords, TOTP codes, and tokens to the attacker-controlled server, potentially granting unauthorized access to the victim's Vikunja account.
Vikunja versions before 2.7.1 have a prompt injection vulnerability that allows project members with write access to inject instructions via project titles or task content. These inputs are rendered verbatim inside the coding agent's instruction block, enabling attackers to influence the agent's behavior on session start and pre-compact events. This can cause unauthorized modifications to developer checkouts or execution of commands.
Vikunja versions before 2.7.1 have a protocol downgrade vulnerability in the serverCandidates component. This flaw allows on-path attackers to force the use of plain HTTP instead of HTTPS during initialization and login. By blocking HTTPS ports, attackers can cause the client to accept unverified HTTP responses, enabling interception of sensitive data such as passwords, TOTP codes, and bot tokens. The vulnerability is fixed in version 2.7.1.
Vikunja versions before 2.7.1 contain a symlink following vulnerability that allows attackers to write files outside the intended repository by using crafted symbolic links. This can lead to overwriting user files such as ~/.bashrc or installing hooks and plugins globally. The vulnerability is fixed in version 2.7.1.
Vikunja versions 2.0.0 up to but not including 2.7.1 are vulnerable to a Cross-Site Request Forgery (CSRF) attack on the /api/v2 login and OIDC callback endpoints. These endpoints accept form-encoded bodies without verifying CSRF tokens or Origin headers, allowing an attacker with an account to trick a victim into switching to the attacker's account. This can expose data the victim subsequently adds. The vulnerability is fixed in version 2.7.1.
Vikunja versions prior to 2.7.1 have a cross-site request forgery (CSRF) vulnerability in their OAuth migration callbacks for Todoist, Microsoft To Do, and Trello. This flaw allows attackers to import data into a victim's account by exploiting missing state parameter binding in the OAuth flow. Attackers can authorize their own accounts and trick logged-in victims into triggering the migration callback, causing attacker-chosen projects and tasks with phishing links to be imported automatically.
Vikunja versions from 2.4.0 up to but not including 2.7.1 have an authentication bypass vulnerability. This flaw allows disabled or locked users to maintain access via API tokens associated with their bot users. The vulnerability arises because token validation only verifies the bot user's status, not the parent user's status. Attackers with pre-existing bot tokens can continue to read and modify shared projects through multiple interfaces including the HTTP API, CalDAV, MCP, and feeds.
The ARTEX attack involves an agentic AI framework that autonomously searches public GitHub repositories for exposed secrets such as OAuth application credentials. Once found, these secrets are abused to access and move laterally within Azure and Microsoft 365 cloud environments, sometimes exfiltrating sensitive data within minutes. ARTEX was publicly available on GitHub until early October 2026 and is now actively used by threat actors in opportunistic campaigns targeting cryptocurrency wallets, gaming accounts, and code-signing certificates. The attack leverages automation and AI to operate at machine speed, significantly accelerating credential abuse and cloud tenant compromise.
This content is a detailed technical research and analysis of the PHP2Shell vulnerabilities, focusing on memory corruption issues such as use-after-free (UAF) and stale pointers within PHP internals. The author provides an in-depth walkthrough from basic memory concepts to the actual exploit code, aiming to clarify how these vulnerabilities lead to remote code execution (RCE). The research is based on personal experiments and source code analysis rather than an official advisory. No specific affected versions or patch information are provided.
The Subscribe2 – Form, Email Subscribers & Newsletters plugin for WordPress contains a Stored Cross-Site Scripting (XSS) vulnerability in the 'data[ip]' parameter. This affects all versions up to and including 10.47. The flaw arises from insufficient input sanitization and output escaping, allowing unauthenticated attackers to inject malicious scripts that execute when a user views the affected page.
The 'twitch-stream-py' package on PyPI was found to be a malicious clone of a legitimate package. It contains obfuscated code that downloads and executes a remote script with malicious intent. The remote script was not accessible during analysis. The campaign was identified by Threx-code and is categorized as malware due to its behavior resembling infostealers.
The pyuserox package version 0.0.2 on PyPI was found to silently download and execute remote malicious code during import. The remote code was no longer accessible during analysis, but the package's behavior indicates clear malicious intent. This campaign was discovered by Threx-code and is categorized as malware due to its capability to execute unauthorized remote scripts.
The pyuserfox package version 0.0.2 on PyPI was found to silently download and execute remote malicious code during import. The remote code is no longer accessible, but the package's behavior indicates clear malicious intent. This campaign was discovered by Threx-code and is categorized as malware.
The fabdeploycheck package on PyPI contains malicious code that downloads a remote configuration and executes commands under attacker control. It establishes a reverse shell, allowing full command execution on the infected host. Additional capabilities include arbitrary file reads, process execution, TCP probing, and HTTP requests with stolen tokens. The package disables TLS verification when fetching commands, enabling man-in-the-middle attacks. It also kills processes related to previous implant versions to maintain persistence. This behavior constitutes a backdoor implant with data exfiltration and remote control capabilities.
Starting October 7, 2026, attackers exploited two critical vulnerabilities in AhsayCBS backup utility (CVE-2026-105133 and CVE-2026-105134) to perform unauthenticated remote code execution. They chained these flaws to bypass authentication and execute code with SYSTEM privileges. Post-exploitation, attackers deployed JSP webshells, installed XMRig cryptocurrency miners disguised as Microsoft Edge processes, and used an AI-assisted PowerShell script to hide mining activity by terminating Task Manager. Persistence was maintained via a fake Windows service named MicrosoftEdgeUpdateSvc. Additionally, a vulnerable kernel driver (WinRing0x64.sys) was exploited for kernel-level hardware access to optimize mining. All AhsayCBS versions through 10.3.4 are affected, with managed service providers primarily targeted.
MobilityDB versions up to 1.3.0 contain a vulnerability in the WKB deserialization logic where a signed to unsigned conversion error allows unprivileged users to cause a denial-of-service by crashing the PostgreSQL backend. This occurs due to a negative length field in a crafted WKB payload that bypasses bounds checks and leads to out-of-bounds memory reads during memcpy operations.
CVE-2026-76875 is a use-after-free vulnerability in PyPy's pyexpat module affecting versions before 3.11.16 and 3.12.14. It occurs when an application creates external-entity sub-parsers without retaining a reference to the parent parser, allowing memory corruption via crafted XML documents.
CVE-2026-93990 is a high-severity vulnerability in libexpat versions before 2.8.5 where improper handling of UTF-16 encoding allows malformed sequences with lone high surrogates to be accepted. This can cause unpaired surrogates to be passed to applications or silent replacement of input characters, potentially impacting XML processing.