Agent Skills in the Wild: An Empirical Study of Security Vulnerabilities at Scale
Jan 15, 2026·,,,,
,,·
1 min read
Yi Liu
Weizhe Wang
Ruitao Feng
Yao Zhang
Guangquan Xu
Gelei Deng
Yuekang Li
Leo Zhang
Abstract
This empirical study analyzes security vulnerabilities in AI-agent skills at scale. It collects skills from major marketplaces, applies SkillScan to detect vulnerable and malicious patterns, and builds a taxonomy spanning prompt injection, data exfiltration, privilege escalation, and supply-chain risks.
Type
Publication
arXiv preprint arXiv:2601.10338
This work studies agent skills as an emerging software supply-chain attack surface and provides empirical evidence for capability-based permissions and mandatory security vetting of skill ecosystems.
