update-docs

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The skill explicitly instructs subagents to ingest full document contents and to list or preserve specific sections/paragraphs (and include "Preserved sections" in summaries and todo entries), which can force the model to reproduce verbatim text from docs — including any embedded secrets — even though it doesn't explicitly ask for credentials.