prompt-analysis

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The skill explicitly tells subagents to read the full messages JSON (which contains complete user/AI conversations) and to return or store evidence from those messages (e.g., failure_analysis, notes, or SQL updates), meaning the LLM will likely see and may be asked to emit verbatim secrets if they appear in prompts/responses.