humanize

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The skill mandates preserving data and passing code/structured blocks unchanged, so if the input contains API keys, tokens, or passwords the LLM is required to reproduce them verbatim, creating an exfiltration risk even though it doesn't explicitly request secrets.