clean-codelisted

# Code Cleanup — Orchestrator *Productivity — Multi-agent orchestration. Audits and refactors existing code in-place for readability, maintainability, and dead-code removal — without changing behavior. Produces a dated cleanup record.* **Core Question:** "Is this change purely structural with zero behavioral impact?" > Why this skill exists, methodology, principles, when NOT to refactor, history: [`references/playbook.md`](references/playbook.md) [PLAYBOOK]. ## When To Use - Codebase has accumulated dead code, AI slop, unused dependencies, or production-waste assets. - After major feature additions, before release milestones, when test runtime grows. - When onboarding new team members and structural cruft slows them down. - Standalone — no upstream gate required. ## When NOT To Use - Cleanup is mixed with a feature change (separate commits — always). - No test coverage AND behavior-preservation matters (write tests first; see [`references/anti-patterns.md`](references/anti-patterns.md) [ANTI-PATTERN] "When NOT to refactor"). - Pre-existing test/build failures unrelated to cleanup (BLOCKED until baseline is green). - Code that won't change again — if nobody will read or modify it, the investment doesn't pay off. - Duplication is the same **operational mechanics** repeated across 2+ callers (identical SDK setup, retry loop, I/O plumbing) → use `/extract-service`. `clean-code` removes dead code and slop; it does not design service boundaries. ## Critical Gates (The 5 Gol