vague

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly records and later outputs the user's "original requirement" verbatim (in the .claude state file and the "Before (Original)" summary), so any secrets included in the requirement would be reproduced by the LLM, creating an exfiltration risk.