BlueTeamSec briefing for Thursday the 19th of November 2026, drawn from what the community was sharing on infosec.pub. Four stories today, starting with a nasty one from the CISA catalogue.
CISA's Known Exploited Vulnerabilities catalogue has added CVE-2026-90436, a critical unauthenticated remote code execution flaw in Apache Axis 1.x. That's an end-of-life library embedded in products like Atlassian Confluence Data Center and N-able N-central. A crafted SOAP fault triggers unsafe deserialization before authentication even runs, there's public proof-of-concept code, and it's being exploited against internet-facing apps. If any of these were exposed, the question is less 'have we patched' and more 'what happened before we did', so look for odd child processes and outbound connections from application servers. We touched on this one in the KEV alerts on the 18th, and it also came up in a podcast episode on the 17th.
Team Cymru is pointing to CrowdStrike's reporting on China-nexus espionage against the global financial sector in the first half of 2026. The activity leans on shared infrastructure: ShadowPad command servers, Vshell servers and networks of relay boxes to stay hidden. Targets include central banks, national payment gateways and big commercial banks, mostly across the Americas, the Middle East and Southeast Asia. Worth a read if you work in finance, or if you just like your threat intel with a long time horizon.
MalwareMustDie has released a detection pack for ARTEX, the open-source, LLM-driven penetration testing framework that attackers are reportedly misusing to squeeze attack lifecycles into hours. It ships in a pile of formats, including Sigma, KQL, Suricata and YARA, and leans on behavioural detection because the agents' infrastructure is ephemeral. That's a sensible choice, since chasing static indicators here would be a hobby rather than a defence. It follows Cisco Talos's ARTEX piece on the 16th, and a Cisco-style 'AI changes everything' claim at least has some linked incidents behind it this time.
NTT Security Japan reports that WaterPlum is targeting developers with a fake job-application coding challenge, delivered as a malicious Git repository. The trick is a crafted Git config that abuses the clean filter, so simply running git status, or an IDE doing its background thing, executes the payload. That drops OtterCandy, and the latest version also deploys NetSupport Manager. Worth flagging to your developers, who are probably the last people to expect a job test to run code just by being looked at.
And that's the briefing for today. Where we've linked activity to a particular country or group, that was the original authors' assessment rather than ours, and the analysis was automated. Thanks for listening.