plan-0-v2-constitution

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). This prompt explicitly directs the agent to extract and preserve "filled placeholder" values and include them in JSON outputs and updated files, which forces the model to read and potentially emit secrets or API keys verbatim, creating an exfiltration risk.