review-change

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The skill ingests full git diffs and untracked-file contents (synthesized "all-additions") into agent prompts without any enforced redaction rules, so any secrets present in those files will be seen by the LLM and could be echoed verbatim in findings — creating an exfiltration risk.