polish

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt requires reading and passing full diffs and exact file lines to review agents and to include code snippets/line references in the report, which can force the LLM to reproduce secret values (hardcoded keys, tokens, passwords) verbatim in its output.