babysit
Warn
Audited by Socket on Apr 1, 2026
1 alert found:
SecuritySecuritySKILL.md
MEDIUMSecurityMEDIUM
SKILL.md
SUSPICIOUS. The core babysitter orchestration behavior mostly matches the stated purpose, and the main SDK install path is plausibly legitimate. However, the skill is high-privilege, directs the agent to install additional skills/subagents, and mixes untrusted web/repo research with file writes and command execution, creating meaningful transitive-trust and prompt-injection risk even without clear exfiltration behavior.
Confidence: 84%Severity: 71%
Audit Metadata