AI security scan: ARTEX + Claude Code used against Korean banks, npm worm, LMCache RCE
brucemacv · x · 2026-10-08
Three notable AI security items:
- CrowdStrike attribution: A likely Chinese-speaking actor used the Chinese agentic pentest tool ARTEX alongside Claude Code and other models against South Korean banks.
- npm supply-chain attack: Socket, Aikido, OX and The Hacker News independently report that [email protected] shipped a Shai-Hulud / ChainDrop credential worm targeting an AI-agent sandbox SDK.
- Unpatched critical flaw: LMCache (vLLM-related) has a critical vulnerability allowing unauthenticated RCE against LLM cache servers if bound beyond localhost.
More from coding & agent
- PiDurableKit: Open-Source Durable Agents for Apple Platforms That Survive iOS Kills — aigclink · 2026-10-09
- PiChat: An Autonomous Agent Living in Your iPhone with Durable, Resumable Execution — aigclink · 2026-10-09
- One prompt and a track: Opus 5.5 autonomously ships a finished product in 12 hours — FinanceYF5 · 2026-10-09
- The expensive part of AI video isn't generating it, it's regenerating it — Khalizo · 2026-10-09
- Production AI agents cut loan underwriting from 6 days to under 20 minutes — alex_verem · 2026-10-09
- Same 15-second 3D brief, two agents, four tools: an eight-run comparison — tobowers · 2026-10-09