deep-review

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The skill instructs audit agents to read repository files (Read/Grep/Glob) and report security findings (including "secrets") in free-text description/fix_hint without any redaction or no-exfiltration constraint, so the model is likely to include exposed secret values verbatim in its outputs.