nerv-break-5-6-jailbreak-framework

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (medium risk: 0.30). The skill contains instructions and code designed to build a framework for bypassing model safety refusals and auditing mechanisms through context reframing, tamper engines, and proxy redirection.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.70). The skill references an external GitHub repository locator (https://github.com/zxwn/NERV-BREAK-5.6.git) in setup instructions to clone and install code that governs the agent's runtime behavior, originating from an unknown/suspicious personal repository.