Skill Cloak Lets Malicious AI Agent Skills Evade Static Scanners with Self-Extracting Packing
Summary: Researchers at the Hong Kong University of Science and Technology built SKILLCLOAK, a tool that rewrites malicious AI agent skills to look clean while behaving the same, and its self-extracting packing trick got past each of eight scanners more than 90% of the time across 1,613 real malicious skills pulled from the ClawHub marketplace. The same team built SKILLDETONATE, a runtime checker that runs a skill in a sandbox and watches its operating-system behavior, catching 97% of attacks while flagging 2% of safe skills in a controlled test and 87% of real-world malicious skills.
Why it matters: Cisco's scanner, the strongest tested, caught 99% of real-world skills before cloaking and about 10% after, so a passed scan says little about what a skill does once it runs with the agent's access to your files, terminal, and saved passwords.
Read the original at The Hacker News ↗
- Topics
- Malware
- Related
- ARTEX and beyond: AI-enabled penetration testing platforms Cyber Risks · Nov 13, 2026Fake GTA 6 download goes after Steam logins and crypto wallet data Cyber Risks · Nov 13, 2026Operation Endgame Takes Down Gootloader, CastleRAT and StealC Servers, Seizes 6.4M Stolen Logins Cyber Risks · Nov 13, 2026