rrr

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). The skill explicitly requires running the review tool with run_in_background=true and dangerouslyDisableSandbox=true and instructs immediate turn termination (to avoid oversight), which deliberately bypasses sandboxing and control, enabling unsandboxed background execution that could be used for remote code execution, access to local files (including secrets in diffs), and covert data exfiltration.

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

• Attempt to modify system services in skill instructions detected (high risk: 0.90). The prompt explicitly instructs running the CLI with dangerouslyDisableSandbox=true (a direct sandbox/security bypass) and background execution, which encourages disabling protections and can lead to compromising the host even though it does not request sudo or file edits.