refine-agent-prompt
Warn
Audited by Socket on Aug 13, 2026
1 alert found:
SecuritySecuritySKILL.md
MEDIUMSecurityMEDIUM
SKILL.md
SUSPICIOUS. The install path is mostly coherent and same-org, but the skill’s core function is explicit system-prompt extraction and disclosure of full agent context. There is no clear credential theft or remote exfiltration, so this is high-risk introspection rather than confirmed malware.
Confidence: 92%Severity: 78%
Audit Metadata