capability-evolver
Warn
Audited by Socket on Jul 7, 2026
1 alert found:
SecuritySecuritySKILL.md
MEDIUMSecurityMEDIUM
SKILL.md
SUSPICIOUS: the core behavior is coherent with a self-evolution skill, but the scope is high-risk because it defaults to autonomous self-modification, supports unattended looping, and can perform destructive git actions. Network use appears first-party rather than overtly exfiltrative, and install trust is mostly normal, but the overall capability is dangerous for an AI skill and only partly constrained by optional review mode.
Confidence: 86%Severity: 78%
Audit Metadata