skill-crafting
Warn
Audited by Socket on Oct 12, 2026
1 alert found:
SecuritySecuritySKILL.md
MEDIUMSecurityMEDIUM
SKILL.md
SUSPICIOUS. The skill's main behavior is consistent with a local eval runner, and there is no evidence of hidden network exfiltration or overt malware. However, it deliberately executes shell fixture commands and an unverifiable local adapter, and it can forward environment variables to that adapter; these capabilities are powerful and should only be used with trusted eval specs and trusted local adapters.
Confidence: 90%Severity: 72%
Audit Metadata