no-sloplisted

# /no-slop One coordinator for public-facing writing and interfaces. It chooses the appropriate source skills, runs them in a deliberate order, applies the user's voice and taste afterward, and requires a final review. Run this sequence: ``` 1. General cleanup copied source skills: /humanizer + /stop-slop for writing, or /taste + /impeccable for interfaces 2. Personal voice the user's own rules, samples, exceptions, and corrections 3. Final review reread the finished work, find what remains, and fix it ``` The coordination is the main value. Using one source skill alone is valid for a narrow job, but it leaves selection, ordering, updates, personal corrections, and final review to the user or agent each time. Keep two improvement paths separate: `scripts/update.mjs` refreshes only copied source files after review, while approved feedback is saved under `voice/` for later work. Never let a source update overwrite the personal path. **Use agent-agnostic language.** In these instructions, "the agent" means whatever LLM-powered agent is running the skill. The core Markdown workflow and the prompt printed by `node scripts/hook.mjs prompt` are portable to systems that can load Agent Skills or instruction folders. Hook configuration is not portable because each agent uses a different format. Name Claude Code only when discussing its exact skill path, settings file, or included hook adapter. **The voice wins.** When an upstream file and a `voi