Briefs

The Kirki plugin for WordPress is vulnerable to arbitrary file deletion via CVE-2026-8073 due to insufficient file path validation and a missing capability check in the 'downloadZIP' function, allowing unauthenticated attackers to delete files within the WordPress uploads directory.

TeamPCP compromised the PyPi package durabletask (versions 1.4.1, 1.4.2, and 1.4.3), stealing credentials for AWS, Azure, GCP, K8s, and Vault, brute-forcing passwords from password managers, and exfiltrating shell history before propagating to up to 5 targets via AWS SSM and Kubernetes.

Windmill versions prior to 1.703.2 are vulnerable to incorrect default permissions in the nsjail sandbox configuration, allowing authenticated users to inject malicious entries into critical system files, leading to potential privilege escalation and man-in-the-middle attacks.

The Shai-Hulud campaign is back and targets maintainer accounts to publish malicious code directly into the software supply chain via npm, recently hitting the Ant Design (AntV) ecosystem and potentially exposing downstream developers to credential theft and remote code execution.

Budibase is vulnerable to persistent stored XSS (CVE-2026-46426) due to unrestricted file upload of active content by authenticated users, leading to potential session cookie theft and account takeover.

HPE published a security advisory (AV26-477) addressing a critical vulnerability in HPE Unified OSS Console (UOC) version 3.1.20 and prior, potentially leading to unauthorized access and control of network operations.

An application that passes an overlong ciphertext buffer to `libcrux_chacha20poly1305::encrypt` or `libcrux_chacha20poly1305::xchacha20_poly1305::encrypt` can experience a panic, leading to a crash if the buffer length is attacker-controlled, affecting libcrux-chacha20poly1305 versions prior to 0.0.8.

The AVX2 implementation of ML-DSA verification in libcrux-ml-dsa mishandles an edge case in the `use_hint` function, potentially allowing an attacker to craft an invalid signature that is accepted by the verifier if the AVX2 implementation is used.

The Motorola Factory Test component (com.motorola.motocit) contains an improper authentication vulnerability, allowing a local attacker to bypass permission checks and access protected device settings by leveraging a writable file descriptor in external storage to open a TCP server.

Composer leaks GitHub OAuth tokens in GitHub Actions logs if they do not match the expected format due to a validation regex, leading to potential unauthorized access.

ZKTeco CCTV cameras are vulnerable to authentication bypass due to an undocumented configuration export port that does not require authentication and exposes critical information about the camera, such as open services and account credentials, as tracked by CVE-2026-8598.

Multiple vulnerabilities exist in ScadaBR version 1.2.0, including CVE-2026-8602, CVE-2026-8603, CVE-2026-8604, and CVE-2026-8605, which could allow for unauthenticated remote code execution.

A path traversal vulnerability (CVE-2025-3465) in ABB CoreSense HM and CoreSense M10 allows unauthenticated local users to access restricted directories, potentially leading to system compromise and information exposure; patch to CoreSense™ HM v2.3.4 and CoreSense™ M10 v1.4.1.31.

A cross-site scripting vulnerability, CVE-2026-4293, exists in multiple Kieback & Peter DDC Building Controllers that could allow an attacker to take control of the victim's browser.

Multiple vulnerabilities have been disclosed in SonicWall Gen6 and Gen7 firewalls, SonicOS, and NSv that can be exploited for authentication bypass, remote code execution, and privilege escalation, specifically CVE-2024-40762, CVE-2024-53704, CVE-2024-53705, and CVE-2024-53706; a proof of concept exploit is available for CVE-2024-53704, which, if exploited, can lead to internal network access and further attacks, including ransomware deployment.

Microsoft disrupted Fox Tempest, a threat actor running a malware-signing-as-a-service (MSaaS) that abuses Microsoft Artifact Signing to generate short-lived code-signing certificates used to sign malware disguised as legitimate software, delivering ransomware and various information stealers to victims across multiple sectors.

Argo CD is vulnerable to stored cross-site scripting (XSS) via manipulated application link annotations, allowing a low-privileged user to execute arbitrary JavaScript in a higher-privileged user's session, leading to privilege escalation.

Mailpit is vulnerable to an unauthenticated remote memory-exhaustion denial-of-service attack due to missing size limits on incoming SMTP DATA and HTTP requests, leading to unbounded memory and disk growth, potentially crashing the application.

The auth-fetch-mcp package is vulnerable to server-side request forgery (SSRF) and disk exfiltration due to unvalidated URLs in the `download_media` and `auth_fetch` tools, allowing an attacker to fetch internal resources, cloud metadata, or loopback addresses, potentially leading to credential theft, internal service enumeration, and sensitive information disclosure.

The `pullArtifact` methods in `Registry` and `OCILayout` use the `org.opencontainers.image.title` annotation from a pulled manifest as a filename, resolving it against the caller supplied output directory without normalization or a containment check, allowing a manifest publisher to write blobs outside of the intended target directory.

A malicious version of the guardrails-ai package (0.10.1) was published to PyPI on May 11, 2026, advising users who installed this version to downgrade and treat the host as potentially compromised, rotating credentials and auditing GitHub accounts, with Snowglobe and Guardrails Hub API keys being invalidated on May 13, 2026.

A path traversal vulnerability exists in zrok copy (CVE-2026-45576) where an attacker-controlled WebDAV or zrok drive can write files outside the destination root by manipulating the DAV `href` response.

NGINX JavaScript is vulnerable to a heap buffer overflow (CVE-2026-8711) when the js_fetch_proxy directive is configured with client-controlled variables and ngx.fetch(), allowing unauthenticated attackers to cause worker process restarts or, with ASLR disabled, code execution via crafted HTTP requests.

Funnel Builder for WooCommerce Checkout versions prior to 3.15.0.3 contains a missing authorization vulnerability in the public checkout endpoint that allows unauthenticated attackers to invoke internal methods and inject malicious JavaScript, impacting checkout page visitors.

HestiaCP versions 1.2.0 through 1.9.4 are vulnerable to IP spoofing (CVE-2026-43634), allowing unauthenticated remote attackers to bypass authentication security controls by manipulating the CF-Connecting-IP HTTP header to circumvent fail2ban, bypass IP allowlists, and poison authentication logs.

The zrok Python SDK `ProxyShare` is vulnerable to server-side request forgery (SSRF) via CVE-2026-45568. When a user sends a request with an absolute URL in the path, the Flask handler passes that path to `urllib.parse.urljoin`, which replaces the configured target host with the user-supplied host, causing the proxy to send the request to an attacker-chosen URL.

HAXcms is vulnerable to stored XSS and exposes authentication tokens in the `/system/api/connectionSettings` endpoint, allowing an attacker to perform cross-tenant account takeover by injecting malicious JavaScript to steal the `jwt`, `user_token`, `site_token`, and `appstore_token`.

Multiple functions in open-apis conduct substring-only matching to validate hostnames, allowing an attacker to perform Server-Side Request Forgery (SSRF) and capture authentication credentials by redirecting requests to an attacker-controlled endpoint.

The HAXcms Node.js backend contains two cryptographic implementation errors in the `hmacBase64()` function that allow an unauthenticated attacker to extract the system’s private signing key and forge arbitrary admin-level JSON Web Tokens (JWTs) allowing them to get full admin access with a single HTTP request.

HAXcms is vulnerable to Server-Side Request Forgery (SSRF) via the createSite endpoint, allowing an authenticated user to supply arbitrary URLs or local file paths, which are fetched server-side without validation and written to a web-accessible directory, enabling arbitrary file read, internal network access, and cloud credential exposure; this vulnerability is tracked as CVE-2026-46393.