sandbox-escape

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). This content contains explicit, actionable exploit payloads and PoCs for sandbox escapes (node:vm constructor-chain, Python class-hierarchy) that enable remote code execution, access to process.env (data/credential exfiltration), and arbitrary command execution—clear malicious/enabling behavior.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 1.00). Because it explicitly provides exploit code and step-by-step instructions to escape sandboxes and achieve remote code execution on the host (e.g., using the constructor chain to access process and child_process.execSync), it directly encourages compromising the machine's state.