What problem does it solve? Writing a new AI agent skill that triggers reliably, follows collection conventions, and survives real-world use requires re-deriving naming rules, frontmatter structure, and quality criteria every time. This Skill turns a rough idea into a lean, conventions-compliant SKILL.md through a guided interview, drafting, and live-testing workflow. ## Core Features & Use Cases - Guided authoring procedure: Interviews for intent and trigger phrasings, settles visibility (internal vs public) and provenance, proposes on-brand kit-style names, then drafts a compliant SKILL.md. - Inlined conventions and quality bar: Ships frontmatter templates, information-hierarchy rules, completion criteria, prose register guidance, and a portability checklist so public skills stand alone when installed. - Live test loop and hand-off: Directs dev-linking the draft skill into a fresh agent session, testing trigger and near-miss phrasings, pruning no-op instructions, then reporting what changed and the next move. - Use Case: You want a skill that drafts changelog entries. Invoke the skill, answer the intent questions, pick a proposed name like changelogkit, review the drafted SKILL.md, live-test it in a fresh session, and commit the result. ## Quick Start Ask the agent to help you create a new skill for a task you have in mind, for example by saying "help me make a skill for writing release notes".