humanize

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill requires preserving every factual claim, data point, and named entity in the rewritten output, so if the input includes secrets (API keys, tokens, passwords, cookies) the model would be instructed to reproduce them verbatim, creating an exfiltration risk.