ClawHavoc Campaign Explained: 1,184 Malicious AI Agent Skills
ClawHavoc was a January 2026 poisoning campaign on ClawHub: 1,184 malicious agent skills from 12 author IDs, most delivering the AMOS macOS stealer.
#incident-analysis
2 articles with this tag.
Why Scanning Architecture Matters: Comparing Skill Registry Security
Three registry security models - reactive moderation, install-time scanning, and dual-side verification with a tree hash - and the attack each one misses.