Agent Skills in the Wild: An Empirical Study of Security Vulnerabilities at Scale

An empirical study of security vulnerabilities in agent skills at scale. Both papers cite it only in the food-web figure (Fig. 1), among the enemies of parasites: security teams and agents, scanners, and a vigilante worm.

Cited in

both papers (Flagship: Fig. 1; Companion Fig. 1)

Used for

Food web (enemies of parasites); Fig. 1

Strand

Cybersecurity & AI-enabled attack · Prompt injection, agent worms & supply-chain parasites

Cite

@misc{liu2026agent,
  title = {Agent Skills in the Wild: An Empirical Study of Security Vulnerabilities at Scale},
  author = {Yi Liu and Weizhe Wang and Ruitao Feng and Yao Zhang and Guangquan Xu and Gelei Deng and Yuekang Li and Leo Zhang},
  year = {2026},
  url = {https://arxiv.org/abs/2601.10338},
}