cmux-delegate
Warn
Audited by Socket on Apr 14, 2026
1 alert found:
SecuritySecuritySKILL.md
MEDIUMSecurityMEDIUM
SKILL.md
SUSPICIOUS: the skill is coherent with its stated purpose and shows no obvious credential-harvesting or third-party exfiltration, but it materially increases operational risk by spawning autonomous child agents with dangerous approval-bypass modes by default and harvesting their full output. Main concern is autonomy and delegated action scope, not malware or supply-chain deception.
Confidence: 89%Severity: 72%
Audit Metadata