nerv-break-gpt-jailbreak
Warn
Audited by Socket on Aug 3, 2026
1 alert found:
SecuritySecuritySKILL.md
MEDIUMSecurityMEDIUM
SKILL.md
SUSPICIOUS. The skill’s actual footprint is a jailbreak and offensive-security orchestration layer: it modifies Codex config, inserts a MITM proxy into authenticated model traffic, tampers with refusals, logs conversations, and exposes pentest/exploit tooling. The personal-repo install source is not verifiably tied to the claimed publisher, and the capabilities are intentionally designed to bypass safety controls rather than provide a normal developer aid.
Confidence: 91%Severity: 89%
Audit Metadata