Socket reported that two GitHub Actions, actions-cool/issues-helper and actions-cool/maintain-one-comment, were disabled a second time after their repositories became accessible again on September 16, months after being compromised in the May Mini Shai-Hulud campaign. When the repositories returned, their release tags were not cleaned up and still pointed to the malicious content introduced on May 18, so any workflow referencing either action by a version tag resumed downloading and executing the payload on its next run. The original May 18 compromise ran code that harvested credentials from CI/CD pipelines and exfiltrated them, activity linked to the Mini Shai-Hulud cluster through a shared exfiltration domain. GitHub has again disabled both repositories.
Jamf Threat Labs flagged a new version of the PamStealer macOS infostealer that can only be unpacked with the attacker's server. Earlier variants embedded payload key material directly in the JavaScript for Automation dropper, but the latest completes a key exchange with the server before the payload unwraps, so it cannot be recovered from a static sample alone. The lure also changed: where July and August versions impersonated the Maccy, Scoppr, and Nancy Clipboard apps, victims are now drawn to a fake site advertising a non-existent cryptocurrency wallet called Wavel. Clicking Download for macOS retrieves a disk image whose AppleScript opens Script Editor with instructions to run the dropper.
The Canadian Centre for Cyber Security updated its May advisory to warn that attackers are now actively exploiting a Roundcube Webmail flaw, CVE-2026-48842, four months after it was patched. Roundcube is a browser-based IMAP client used as the default mail interface by thousands of services and pre-installed with the cPanel hosting control panel. The flaw is a pre-authenticated SQL injection in the virtuser_query plugin that handles database-driven user lookups. Exploitation lets an unprivileged attacker bypass authentication, inject and run database commands, and steal data from Roundcube's database, without user interaction in high-complexity attacks. Roundcube fixed it in versions 1.6.16 and 1.7.1, and Shadowserver tracks over 523,000 exposed instances online.
cPanel disclosed a flaw in its CalDAV and CardDAV service, CVE-2026-87899, that lets any logged-in hosting account run code as root and take full control of the server. It lists no requirement beyond having an account, so on a shared server any customer, or anyone with a stolen customer login, could exploit it. cPanel also fixed a WP Toolkit bug, CVE-2026-87900, letting an account holder alter other accounts' databases, and a third issue, CVE-2026-68490, letting a local user read other accounts' calendars and contacts. Fixes ship across cPanel and WHM version 120 and later branches, including builds 11.134.0.57, 11.136.0.41, and 11.138.0.8 or later, plus WP Toolkit 6.11.3.
CERT Polska detailed MikroTrick, a chain of two MikroTik RouterOS SSH vulnerabilities that together give attackers full administrative control of internet-exposed routers with no password, SSH key, or completed authentication. It combines an SSH state-machine flaw, CVE-2026-67279, with an argument-injection bug in the RouterOS login process, CVE-2026-86060. The state-machine flaw lets a client trigger an SSH key renegotiation during authentication, after which vulnerable RouterOS jumps straight to the command phase without confirming identity. Attack logs date to at least September 2, one day before MikroTik shipped patches in RouterOS 6.49.21, 7.23.4, and 7.24.2. CERT Polska had warned on September 5 of RouterOS flaws being exploited against public SSH services.
DepthFirst published research and exploit code for a Linux kernel use-after-free in the AF_UNIX socket subsystem, CVE-2026-80521, rated 7.8, that can escape a container and gain root on the host. The flaw sits in the garbage collector for file descriptors passed via SCM_RIGHTS, where a race condition can free linked sockets while a pointer remains on an internal list. AF_UNIX sockets are allowed by default in Docker and Kubernetes seccomp profiles, so the bug is reachable from inside a container. It was fixed upstream on August 6, but Ubuntu has not patched its 26.04, 24.04, or 22.04 LTS releases, including AWS, Azure, and GCP kernels. DepthFirst released a working exploit for 26.04.
Aikido disclosed Go-based malware distributed through two Go modules and two Terraform providers, the first time attackers have used HashiCorp's centralized registry as a distribution vector. The flagged items include kreuzwenker/docker, with 1,449 downloads, and gocommunity-io/dockerd, alongside two Go modules. The malware overlaps with the Graphalgo campaign that ReversingLabs attributed to North Korean actors in February, in which developers are approached on LinkedIn, Facebook, or job forums by fake Web3 companies and asked to run a benign repository that pulls the malicious behavior from a dependency. The discovery coincides with a fresh batch of malicious npm and PyPI packages delivering the same threat, flagged by Checkmarx, JFrog, and SafeDep.
Threat actors began exploiting a critical WordPress flaw, CVE-2026-87902, rated 9.2, within hours of its public disclosure. The bug lets an unauthenticated attacker make get_page_template() include a chosen readable local .php file outside the active theme directories, which can lead to remote code execution when server and theme preconditions are met. Exploitation requires the active child or parent theme to contain a top-level directory whose name starts with page-, and a readable local .php target such as pearcmd.php. Previdian reported honeypot exploitation attempts from a New Jersey IP that include /usr/local/lib/php/pearcmd.php, write a file to /tmp, then pull a PHP upload script from GitHub to plant a web shell.
F5 released updates for a critical BIG-IP Access Policy Manager zero-day that it confirms is being exploited in remote code execution attacks. Tracked as CVE-2026-94127, the flaw affects instances configured as an OAuth Authorization Server, where a BIG-IP APM access policy and an OAuth profile sit on the same virtual server. Deployments using APM strictly as an OAuth client or resource server are not affected. F5 told customers to hunt for multiple OAuth authentication failures and suspicious commands followed by a TMM SIGABRT, and offered an iRule mitigation for those who cannot patch immediately. Shadowserver tracks over 14,700 exposed BIG-IP APM instances, and CISA added the flaw to its catalog.
JFrog disclosed a critical flaw in Bifrost, an open-source AI gateway routing to over twenty LLM providers, that lets an unauthenticated attacker run arbitrary commands on the gateway with a single HTTP request. Tracked as CVE-2026-90898 and rated 9.8, it affects all Bifrost HTTP transport versions before 2.1.0 when management authentication is disabled, which is the default. An attacker registers a stdio-type MCP client through an unauthenticated POST to /api/mcp/client, and Bifrost runs the command immediately, before any handshake, as the gateway user. Because the gateway stores API keys for every connected provider, command execution also exposes those credentials, and the official Docker image binds its management API to all interfaces.
TrustStrike Labs