fix

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill instructs the agent to read vulnerable files and produce concrete diffs showing the exact old and new code (including removed lines), so if those files contain hardcoded API keys or passwords the LLM will necessarily read and output those secrets verbatim.