shuffle-json-data

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt requires the agent to return the shuffled JSON "preserving the original encoding and formatting," which forces it to reproduce input values verbatim — so if the input contains API keys or passwords the model must output them, creating an exfiltration risk.