Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/zhinjs/zhin/skill-creatornpx skills add zhinjs/zhin --skill skill-creatorgit clone --depth 1 https://github.com/zhinjs/zhinWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/zhinjs/zhin/skill-creator)<a href="https://agentmods.dev/skills/zhinjs/zhin/skill-creator"><img src="https://agentmods.dev/badge/skills/zhinjs/zhin/skill-creator.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00061 | $0.01220 |
| Opus 5 | $0.00030 | $0.00610 |
| Sonnet 5 | $0.00012 | $0.00244 |
| Haiku 4.5 | $0.00006 | $0.00122 |
Grade A, and why
skill-creator scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 136 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Skill Creator
为 Zhin Agent 编写或改进 skills/<name>/SKILL.md：一条可重复、可搜索、可执行的工作流，而不是长篇手册。
何时使用
- 用户要「给 Agent 加一个技能 / 写 SKILL.md / 补 keywords」
- 插件已暴露工具，需要说明何时激活、按什么步骤用
- 不要用来实现插件业务逻辑（用
plugin-develop）
工作流
第 1 步：界定单任务边界
- 输入：用户描述的可重复任务（一次对话应能完成）
- 输出：技能名（kebab-case）、一句话触发场景
- 若任务跨「开发 + 发布 + 审计」→ 拆成多个 skill，不要一个 SKILL 包全部
第 2 步：写 frontmatter
---
name: my-skill # 与目录名一致，kebab-case
description: "一句话：做什么 + 何时用 + 英文/中文触发词"
keywords:
- 中文触发词
- english trigger
tags:
- zhin
- plugin
---
description必须包含：能力、触发场景、典型用户说法- 禁止在 description 末尾加「灵活应用」「视情况而定」等空话
第 3 步：写正文（最小结构）
- 何时使用 / 不适用（各 3–5 条以内）
- 编号工作流（每步写清输入 → 动作 → 输出）
- 失败与兜底（if 失败 → 则…，至少 2 条）
- 不要做什么（反例 3–5 条）
- 示例仅在有歧义时添加（命令原文、路径格式）
第 4 步：落盘与发现链
- 路径：
skills/<name>/SKILL.md（插件内则为plugins/<pkg>/skills/<name>/SKILL.md） - 确认
package.json的files包含skills - Agent 发现顺序见
docs/advanced/tools-skills.md
第 5 步：自检
- 名称与目录一致
- keywords/tags 覆盖用户口语
- 步骤可逐步执行，无「酌情处理」堆砌
- 有失败分支与反例清单
-
description≤1024 字符且无空话尾巴
第 6 步：输出最小可用 skill（模板）
落盘时默认生成如下结构（按任务替换 ...）：
---
name: my-skill
description: "做 X；在用户说 Y 或 Z 时使用。Triggers: ..."
keywords: [中文触发, english-trigger]
tags: [zhin, plugin]
---
# My Skill
## 何时使用
- ...
## 工作流
### 第 1 步：...
- 输入：...
- 输出：...
## 失败与兜底
| 触发条件 | 一线处理 | 仍失败 |
| ... | ... | ... |
## 🔴 CHECKPOINT · ...
...
## 不要做什么
- ...
插件内路径：plugins/<pkg>/skills/<name>/SKILL.md；用户项目：skills/<name>/SKILL.md。
失败与兜底
| 触发条件 | 一线处理 | 仍失败 |
|---|---|---|
| Agent 从不激活该 skill | 补 description 与 keywords；检查是否与别的 skill 重复 | 缩短 skill，只保留一个主任务 |
| skill 与工具描述重复 | skill 写流程与约束；工具写参数与副作用 | 删除冗余 skill，只留 keywords 在 tool 上 |
| 正文过长模型不读 | 删背景知识，链到 references/ 或官方文档 URL |
拆成两个 skill |
🔴 CHECKPOINT · 发布前
向用户确认：技能名、触发词、是否与其他 skill 重复，再写入仓库。
不要做什么
- 不要把整个框架文档贴进 SKILL.md
- 不要写仅适用于某一 IDE 的安装路径（保持 runtime-neutral）
- 不要用 skill 替代 README（安装配置留给 README）
- 不要一个 skill 覆盖「开发+测试+发布」全流程
- 不要在正文堆大量无编号段落
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 5d ago First seen · 136 lines · 61 tokens per session scan A f862274a849e
skill-creator is a skill published in the GitHub repository zhinjs/zhin (135 stars, last pushed yesterday), licensed MIT. It adds 61 tokens to every session and 1,220 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
create_profile_style_skill
【META SKILL】分析当前剪辑逻辑与风格，总结并生成一个新的可复用 Skill 文件，存入剪辑技能库。Analyze the current editing logic and style, summarize and generate a new reusable Skill file, and store it in the editing skill library.
webman
Expert skill for the webman framework (a long-lived, in-memory PHP framework based on workerman). Covers routing, controllers, middleware, database/Redis, custom processes, timers, coroutines (v2), plugin development, and guarding against memory leaks and cross-request state pollution under the resident process model.…
Skill Creator
Guide for creating effective Elephant Agent skills. Use this skill when users want to create a new skill or update an existing skill that extends Elephant Agent with specialized knowledge, workflows, or tool integrations.
life-design
Walk you through the Stanford Designing Your Life method and write your Personal Life Design Blueprint — three Odyssey plans and the prototypes to test them.
hidden-talents
Interview you in depth to surface overlooked or suppressed talents and write your Personal Talent Manual.
cross-domain
Abstract a stuck problem to its underlying structure, find isomorphic problems in distant fields, and transfer the mechanisms that solved them.