simplify

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt instructs the model to fetch and reproduce unified diffs and produce concrete code fixes verbatim, which can require reading and outputting secrets present in the repository (API keys, tokens, passwords) and thus enables exfiltration risk.