self-improving
Warn
Audited by Socket on Sep 21, 2026
1 alert found:
SecuritySecuritySKILL.md
MEDIUMSecurityMEDIUM
SKILL.md
SUSPICIOUS: the skill's stated purpose is coherent, but its real footprint is high-risk because it instructs the agent to install other agent packages from a registry, including community-authored content. PRPM itself appears to be an official project, so this is not confirmed malware, but the transitive installation model, expanded trust surface, and inaccurate 'all searches are local' privacy claim make the skill high security risk.
Confidence: 90%Severity: 76%
Audit Metadata