CVE-2026-84869 is a critical vulnerability in ConnectWise ScreenConnect that allows unauthorized malicious guests in remote sessions to transfer and execute files on MSP technician workstations with elevated privileges. Attackers exploited this flaw in a large-scale campaign targeting 42 MSP servers, using techniques such as worm-like propagation and path traversal to inject startup scripts and deploy rogue clients. The vulnerability enables full compromise of technician workstations, which hold sensitive RMM consoles and customer credentials. The exploitation campaign included the use of deceptive guest sessions and canary files to verify success before payload deployment.
CubeSandbox is an instant, concurrent, secure, and lightweight sandbox for AI agents. Prior to 0.5.1, any authenticated tenant permitted to call POST /sandboxes can supply metadata["host-mount"] with an arbitrary existing host path because the control plane performs no authorization check or allowed-prefix validation for the path. injectHostDirMounts forwards the path to prepareHostDirVolume, where Cubelet uses mount --rbind through the function to perform a recursive bind mount as root and exposes the host subtree to the tenant's MicroVM through virtio-fs, read-write by default. This allows host-root file access and modification, host code execution, access to the container runtime, and compromise of co-resident tenants without a kernel or virtual-machine-monitor flaw. This issue is fixed in version 0.5.1.
A vulnerability in LibreOffice allows it to open web query files (.iqy) that specify a web page to import into a spreadsheet. When such a file is opened via the command line, LibreOffice previously did not apply the same address validation checks as it does for direct command line addresses. This flaw could allow a crafted .iqy file to cause LibreOffice to perform unintended actions, including potentially executing code. The issue is addressed in fixed versions by restricting accepted addresses in .iqy files to only http or https URLs.
LibreOffice Calc allows linking a cell range to an external data source, with the link saved in the document. A vulnerability exists where opening a crafted document can cause a Java database driver to be loaded from a remote location outside the local machine, enabling execution of arbitrary code. This issue arises because the fix for a previous vulnerability (CVE-2026-63277) did not cover this attack vector. The fix requires that Java class path entries be file URLs pointing only to local files.
A command injection vulnerability exists in LibreOffice on Linux and Unix-like systems when using the 'Send > Email Document' feature. The vulnerability arises because the script that launches the mail client allows the shell to interpret the document's file name, enabling execution of arbitrary commands if the file name is crafted maliciously. This issue has been addressed in fixed versions by passing the file name and other parameters as plain arguments to the mail client, preventing shell interpretation.
qbee-agent is the device agent component of the qbee.io IoT device management platform. Prior to 2026.41, the unpackTar extraction logic in app/utils/tar.go validates cleaned archive entry paths with strings.HasPrefix(targetPath, destdirClean) without enforcing a path-separator boundary. When UseContext enables the Docker Compose build context path in app/configuration/bundle_docker_compose.go, a less-privileged archive supplier with delegated configuration and/or file-manager access can provide a sibling-prefix traversal entry that passes the check, causing utils.UnpackTar to create directories and regular files outside the intended context extraction directory with qbee-agent process privileges, despite the supplier lacking authority to write those sibling files. In deployments where only full device or fleet administrators can provide Docker Compose context archives, this behavior is hardening rather than a crossed security boundary. The demonstrated impact is limited to nearby cache or project paths sharing the destination prefix, may affect integrity or availability of qbee-agent-managed Docker Compose state, and does not establish arbitrary absolute-path writes or direct code execution. This issue is fixed in version 2026.41.
A vulnerability affecting multiple software vendors arises from a mismatch between Apple File System (APFS) Unicode normalization and application pathname comparisons. This flaw allows local attackers to bypass file exclusion filters or path locks by using alternative Unicode spellings or case-folded variants, enabling unauthorized file overwrites wherever the process has write permissions. The issue impacts archive tools, packaging systems, and runtimes, potentially causing data corruption, configuration hijacking, or code execution. Several CVEs have been assigned, and patches have been issued by some vendors. The CERT/CC recommends updating affected software to the latest versions and implementing path validation consistent with the filesystem's normalization behavior.
A memory corruption vulnerability exists in VideoLAN VLC 3.0.23. A specially crafted Interplay MVE file can trigger a condition that allows access to a variable that is not explicitly initialized, which under certain conditions can lead to code execution in the context of the current process.
Contentlayer @contentlayer/core through 0.3.4 contains a code injection vulnerability that allows attackers to inject JavaScript by embedding single quotes in content file names. Attackers can add a content file whose name breaks out of the import specifier in makeDataExportFile(), executing injected code when contentlayer dev or a Next.js dev server imports the generated module.
Contentlayer @contentlayer/source-files through 0.3.4 contains a code injection vulnerability that allows attackers who control content files to execute JavaScript by abusing gray-matter's still-enabled JavaScript engine in parseMarkdown(). Attackers can start a Markdown or MDX file with a ---js or ---javascript frontmatter delimiter to run code during build, exposing CI build secrets.
Microsoft jxrlib through 1.1 and the mircomir jxrlib fork before 1.3.1 contain a global buffer over-read vulnerability in DecodeSignificantRun() in image/decode/segdec.c due to an unchecked run length. Attackers can supply a crafted JPEG XR file with an out-of-range iMaxRun value that indexes gSignificantRunBin[] out of bounds, causing memory disclosure or crashes during decoding.
An uncontrolled search path vulnerability (CWE-427) exists in statically linked OpenSSL libraries within Netskope Client (stAgentSvc.exe) and Endpoint DLP (epdlp.exe) on Windows. This flaw allows an authenticated local attacker to execute arbitrary code with SYSTEM privileges by placing a malicious openssl.cnf file on a secondary drive where default Windows permissions permit directory creation. macOS, Linux, and mobile platforms are not affected under default permissions.
Kalmia versions through 0.2.0 have a stored cross-site scripting (XSS) vulnerability that allows read-only authenticated users to upload HTML or SVG files without type restrictions. These files can be accessed unauthenticated via a file retrieval route, enabling execution of JavaScript in the administrator's origin context. This can lead to theft of administrator JWT tokens stored in localStorage.
A vulnerability in Rexroth IndraWorks allows remote code execution when a user opens a specially crafted file. The flaw arises from improper sanitization of path names in the file parser, enabling traversal outside the intended directory and unauthorized file writes. Exploitation requires user interaction and can lead to full system compromise.