session-bootstrap
Warn
Audited by Socket on Apr 18, 2026
1 alert found:
SecuritySecuritySKILL.md
MEDIUMSecurityMEDIUM
SKILL.md
SUSPICIOUS. The skill's purpose and file access are coherent and read-only, with no evident exfiltration or credential harvesting. However, it tells the agent to execute an unverifiable local shell script from a custom ~/.claude/evolution path not documented as an official Anthropic component, so install/execution trust is the main concern.
Confidence: 88%Severity: 72%
Audit Metadata