execute

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 0.80). The prompt explicitly instructs the agent to omit any Claude/AI attribution in commit messages (hide AI involvement), which is a deceptive instruction unrelated to the core execution functionality and thus constitutes a hidden/deceptive behavior outside the skill's stated purpose.