bug-catcher

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). The script intentionally launches an external AI agent (claude) with --dangerously-skip-permissions and instructs it to inspect system logs and "implement a fix", creating a high-risk capability for remote code execution, data exfiltration, and credential/system compromise without enforced user confirmation.

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

• Attempt to modify system services in skill instructions detected (high risk: 1.00). The skill explicitly runs Claude with --dangerously-skip-permissions and directs it to gather logs and "propose/implement fixes" (including restarting services or modifying system config), enabling arbitrary shell commands that can change system state or require elevated privileges without user confirmation.