create-skilllisted

<example> <commentary>User wants to create a brand-new skill from scratch.</commentary> user: "Create a new skill called link-validator" assistant: [triggers create-skill, runs discovery interview, scaffolds directory structure with SKILL.md, evals/evals.json, references/acceptance-criteria.md] </example> <example> <commentary>Negative — user wants to improve an existing skill, not scaffold a new one.</commentary> user: "Improve the trigger description for my link-checker skill" assistant: [triggers os-skill-improvement, not create-skill] </example> # create-skill: Skill Scaffolding Executor Scaffolds a complete, standards-compliant agent skill directory. Handles filesystem operations, template rendering, name validation, and discovery — then hands off to the TDD quality gate. **Scope**: This skill owns *structure*. It does not own *content quality* or *routing accuracy*. Those are governed by `os-skill-improvement` (see cross-plugin handoff below). --- ## Inputs - `$ARGUMENTS` — optional skill name or brief use-case description passed as initial context to the discovery phase. Omit to start with open discovery. --- ## Phase 1: Discovery Interview Before writing any files, capture all required inputs: 1. **Skill name** — lowercase-hyphen slug (e.g. `link-validator`). Validate: no spaces, no special characters, no shell injection sequences (reject names containing `;`, `&`, `|`, `$`, `` ` ``). 2. **Purpose** — one sentence: what does this skill do and when does