refactor

SUSPICIOUS: the core refactoring/edit/test behavior is broadly consistent with the stated purpose, but the skill is high-footprint for an agent skill: it performs autonomous edits and commits, persists team config, and instructs installation/loading of another skill, creating transitive trust risk. No clear credential theft, exfiltration endpoint, or malicious installer chain is present.