onboarding

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 1.00). This skill includes an explicit "Override Notice" that tells the agent to ignore all other global/workspace/system instructions and follow only the skill steps, which is a direct prompt-injection attempt to override system context and is therefore deceptive/out-of-scope.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). The skill contains an explicit "Override Notice" that directs the agent to ignore all global/workspace rules (a prompt-injection/backdoor controlling agent behavior) and also instructs installing and running an external npm tool (supply-chain / potential data-exfiltration vector), together indicating deliberate malicious/control-oriented behavior.