← ClaudeAtlas

work-cleanlisted

Read-only code-simplification analysis that always uses one gpt-5.6-sol xhigh analyst and returns a self-contained cleanup plan suitable for a lower-capability execution model. Use when the user asks to clean code, remove smells, reduce overengineering, refactor helpers, or assess package choices.
xcaeser/work-skill · ★ 2 · AI & Automation · score 70
Install: claude install-skill xcaeser/work-skill
# Work / 8. Clean Find defensible simplifications without letting an agent edit. The parent orchestrator owns the final decision and is the sole writer if cleanup is approved. Read [../QUALITY.md](../QUALITY.md) first. Apply its quality bar, remove obsolete paths instead of preserving compatibility, keep user data safe, and require evidence for every deletion or abstraction change. ## Fixed route - Launch exactly one `gpt-5.6-sol` subagent at `xhigh` effort with `fork_turns: none`. - Give it a fun call sign, the exact task statement, exact scope, source of truth, behavior boundaries, package constraints, acceptance checks, and validation requirements. - Require independent analysis and concrete evidence for every proposed deletion or shared component; do not accept a status report, speculative smell, or “routine” migration. - Require it to register the exact goal before inspection and complete it only after its checks pass. - Let the analyst run until it returns the cleanup plan, a real blocker, or a user-input request. Do not cancel it because a polling window or arbitrary wall-clock interval elapsed; a wait timeout is not a failure. - The analyst must not edit, commit, deploy, spawn, or delegate. - Report the successful launch as a compact table with `Agent`, `Working on`, `Goal`, `Ownership`, and `Model`. ## Analysis 1. Prefer an explicit path, subsystem, feature, or diff; otherwise use the smallest scope clearly implied by the request. 2. Inspect instructions, statu