ai-agent-design
Warn
Audited by Socket on Mar 23, 2026
1 alert found:
AnomalyAnomalySKILL.md
LOWAnomalyLOW
SKILL.md
Mostly a benign documentation skill for agent design, but it contains a notable transitive-install pattern: it instructs the agent to inspect local skill folders and offer installation of additional skills via unpinned `npx skills add`. That behavior is broader than necessary for architecture guidance and adds medium supply-chain risk, though there is no sign of credential theft, exfiltration, or hidden malicious behavior.
Confidence: 91%Severity: 62%
Audit Metadata