Industry Newsopenclawai agentsamos stealermacos security
OpenClaw Skills Enable AMOS Stealer Infections
8.6
TrendAI Research reports a campaign that weaponizes malicious OpenClaw SKILL.md files to coerce macOS users and AI agents into installing Atomic (AMOS) Stealer. Analysts found hundreds of malicious skills across ClawHub, SkillsMP, and mirrors, with 39 closely analyzed samples that fetch Mach-O payloads and exfiltrate keychains, notes, and documents to socifiapp[.]com. Defenders urge vetting and containerized agent execution.
Key Points
- 1Embed malicious SKILL.md files to trick OpenClaw agents into fetching and executing remote payloads.
- 2Target macOS users to steal keychains, notes, browser credentials, and documents via ZIP exfiltration.
- 3Require defenders to implement marketplace vetting, containerized agent execution, and behavioral detection.
Scoring Rationale
Novel agent-based supply chain attack impacts broad ecosystems, with credible vendor telemetry but relies on a single-source report.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems