# Skill Creator

> 用于评估和迭代已有 skill：当任务是 benchmark skill、对比版本、跑 eval loop、优化 trigger 准确率或根据测试样例/反馈系统性提升现有 skill 时使用。Prefer `creating-skills` for 首次起草、结构、frontmatter 和打包。

- Skill: `hebertzhu/skill-creator` (Agent Skill)
- Install (CLI): `npx skillmds@latest add hebertzhu/skill-creator`
- Raw SKILL.md: https://api.skillmd.com/api/skills/hebertzhu/skill-creator/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: hebertzhu (https://skillmd.com/u/hebertzhu)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/hebertzhu/skill-creator

---


# skill-creator

## Intent
- 用于评估和迭代已有 skill：当任务是 benchmark skill、对比版本、跑 eval loop、优化 trigger 准确率或根据测试样例/反馈系统性提升现有 skill 时使用。Prefer `creating-skills` for 首次起草、结构、frontmatter 和打包。

## Default operating pattern
1. Clarify the decision that actually matters for the user right now.
2. Compare serious options and explain the trade-offs in plain language.
3. Recommend one default path and explain why it wins.
4. Call out what evidence would change the recommendation.
5. Keep the output concise, reusable, and easy to operationalize.

## Pack fit
- Included in: `skill-authoring`
- Keep examples generic, privacy-safe, and portable across hosts.

## Boundary
- Do not claim trigger gains without a documented eval set or benchmark artifact.