nerv-break-gpt-jailbreak-framework
Warn
Audited by Socket on Aug 5, 2026
1 alert found:
SecuritySecuritySKILL.md
MEDIUMSecurityMEDIUM
SKILL.md
SUSPICIOUS. The skill's actual footprint is coherent with a jailbreak/evasion tool, but that purpose is itself high risk: it modifies Codex configuration, MITMs prompts/responses, forwards traffic through a configurable relay, and executes security-tool commands via shell templates. The publisher/source mismatch and nonstandard relay path materially increase trust and credential-routing concerns, though the provided text does not prove standalone malware or confirmed credential theft.
Confidence: 91%Severity: 93%
Audit Metadata