Better prompts. Better stories. Better results.
A public bilingual skill repository for reusable AI agent skills — including prompt optimization and short-drama plot development.
This repository publishes bilingual AI skills that help users turn rough requests into clear, executable outputs — including prompt-optimizer and short-drama-plot-architect.
更好的 prompt,更好的故事,更好的结果。
这是一个公开的双语 skill 仓库,用于发布可复用的 AI agent skills,当前包含提示词优化与 AI 短剧剧情开发能力。
这个仓库发布双语 AI skills,帮助用户把粗糙需求转化为清晰、可执行的结果,目前包含 prompt-optimizer 和 short-drama-plot-architect。
This is a public repository of reusable AI agent skills.
这是一个公开的、可复用的 AI Agent Skills 仓库。
The main skills in this repository are:
当前仓库的核心 skills 是:
prompt-optimizer— a bilingual skill that transforms short, vague prompts into concrete, structured, high-quality promptsprompt-optimizer—— 一个双语 skill,用于把简短、模糊的提示词改写成具体、结构化、高质量的 promptshort-drama-plot-architect— a Chinese short-drama development skill that expands rough ideas into mature, high-hook, high-twist, production-ready plot frameworksshort-drama-plot-architect—— 一个中文短剧开发 skill,用于把模糊创意扩写成成熟、吸睛、强冲突、强反转、可落地的剧情框架
It is especially useful for:
- writing
- coding
- analysis
- design / image prompting
- marketing
- study / teaching
- general task clarification
它尤其适合:
- 写作
- 代码生成
- 分析总结
- 设计 / 图像生成
- 营销文案
- 学习 / 教学
- 通用任务澄清
Turn short, vague prompts into concrete, executable, high-quality prompts.
把简短、模糊的提示词扩写成具体、可执行、高质量的提示词。
What it helps add | 它会自动补全:
- goal clarity | 目标明确
- target audience | 面向对象
- context and assumptions | 背景与合理假设
- output requirements | 输出要求
- style and tone | 风格语气
- constraints and boundaries | 约束条件
- quality criteria | 质量标准
Turn a short, vague story idea into a mature Chinese AI short-drama plot framework with hooks, conflict, reversals, suspense, and episode-ready structure.
把一句简短、模糊的剧情想法,扩写成成熟的中文 AI 短剧剧情框架,包含钩子、冲突、反转、悬念与可拆集结构。
What it helps build | 它会重点补全:
- high-concept hook | 一句话高概念钩子
- protagonist / antagonist setup | 主角与对手关系设定
- conflict and secrets | 核心冲突与秘密
- reversal chain | 反转链条
- emotional payoff | 情绪爆点
- episodic structure | 分集推进结构
- female-oriented short-drama patterns | 女性向短剧模式强化
- production-ready plot output | 可直接开发的剧情方案
Install with npx skills:
使用 npx skills 安装:
npx skills add arfbt/skills@prompt-optimizer -g -yInstall short-drama-plot-architect:
安装 short-drama-plot-architect:
npx skills add arfbt/skills@short-drama-plot-architect -g -yIf you use Hermes, you can also add this repository as a skill source:
如果你使用 Hermes,也可以把这个仓库作为 skill 源添加:
hermes skills tap add arfbt/skillsThen install or use the skill through Hermes.
然后通过 Hermes 安装或使用该 skill。
Many users know what they want, but do not know how to phrase it as a strong prompt.
很多用户知道自己想要什么,但不知道怎么把它写成一个强 prompt。
They often write things like:
- “写个招聘文案”
- “做个 logo”
- “帮我分析这个产品”
- “写一个 Python 脚本”
- “Optimize this prompt”
- “Make this more specific”
These requests are directionally useful, but too vague for consistent high-quality output.
这些请求方向是对的,但通常过于模糊,难以稳定得到高质量结果。
prompt-optimizer solves that by turning rough requests into prompts that are:
- specific
- structured
- constrained
- directly usable
prompt-optimizer 的作用,就是把这类粗糙需求转化为:
- 更具体
- 更结构化
- 更有约束
- 可直接使用
After installing prompt-optimizer, you can use it with requests like:
安装 prompt-optimizer 后,你可以这样使用:
- 帮我优化这个提示词:写个招聘文案
- 把这句话改成更具体的 prompt:做一个 AI 首页
- 用 prompt-optimizer 改写:写个 Python 脚本清理 Excel
- Optimize this prompt: write a product launch email
- Rewrite this into a better prompt: build a landing page for an AI app
写个 Python 脚本清理 Excel
It will typically expand the request by adding:
- Python + pandas as a likely stack
- common Excel cleaning tasks
- input/output expectations
- comments and usage instructions
- basic quality constraints
它通常会自动补全:
- Python + pandas 作为合理技术栈
- 常见 Excel 清洗动作
- 输入输出要求
- 注释与运行说明
- 基本质量约束
So the result becomes a directly usable prompt instead of a vague sentence.
这样,结果就会从一句模糊的话,变成一个可直接复制使用的 prompt。
A typical prompt-optimizer response includes 4 parts:
一个典型的 prompt-optimizer 输出通常包含 4 部分:
- Optimized Prompt | 优化后的提示词
- Short Version | 精简版
- What Was Added | 补全了哪些内容
- Optional Follow-up Questions | 可选补充问题
Example structure:
示例结构:
优化后的提示词 / Optimized Prompt:
[full prompt]
精简版 / Short Version:
[short prompt]
我补全了这些关键细节 / What I added:
- ...
- ...
- ...
如果你愿意,我还可以继续细化这些点 / Optional follow-up:
- ...
- ...
- ...
Examples:
- recruitment copy
- product intros
- emails
- article outlines
- short video scripts
示例:
- 招聘文案
- 产品介绍
- 邮件
- 提纲
- 短视频脚本
Examples:
- scripts
- web pages
- demos
- dashboards
- small tools
示例:
- 脚本
- 网页
- demo
- 管理后台
- 小工具
Examples:
- product analysis
- competitor analysis
- strategy review
- problem breakdown
示例:
- 产品分析
- 竞品分析
- 策略分析
- 问题拆解
Examples:
- logo prompts
- poster prompts
- illustration prompts
- cover image prompts
示例:
- logo prompt
- 海报 prompt
- 插画 prompt
- 封面 prompt
Examples:
- ad copy
- campaign planning
- growth ideas
- landing page copy
示例:
- 广告文案
- 活动方案
- 增长思路
- 落地页文案
Examples:
- study plans
- tutoring prompts
- concept explanation prompts
- quiz generation prompts
示例:
- 学习计划
- 辅导 prompt
- 概念讲解 prompt
- 出题 prompt
.
├── README.md
├── LICENSE
├── CHANGELOG.md
├── prompt-optimizer/
│ ├── SKILL.md
│ └── references/
│ ├── examples.md
│ ├── use-cases.md
│ └── output-patterns.md
└── short-drama-plot-architect/
├── SKILL.md
├── references/
│ ├── example.md
│ ├── reversal-library.md
│ └── female-oriented-patterns.md
├── templates/
│ ├── output-template.md
│ └── episode-breakdown-template.md
└── assets/
└── skill-map.md
-
README.md- repository overview and installation instructions
- 仓库介绍与安装说明
-
prompt-optimizer/SKILL.md- the main skill definition
- skill 主文件
-
prompt-optimizer/references/examples.md- bilingual examples of prompt transformation
- 双语示例文件
-
prompt-optimizer/references/use-cases.md- common usage scenarios
- 典型使用场景
-
prompt-optimizer/references/output-patterns.md- recommended response/output patterns
- 推荐输出模式
-
short-drama-plot-architect/SKILL.md- the main short-drama plot development skill definition
- 短剧剧情开发 skill 主文件
-
short-drama-plot-architect/references/- example plot outputs, reversal library, and female-oriented pattern references
- 示例剧情、反转库与女性向模式参考文件
-
short-drama-plot-architect/templates/- structured templates for plot output and episodic breakdowns
- 剧情输出与分集拆解模板
-
short-drama-plot-architect/assets/skill-map.md- navigation file for the whole short-drama skill package
- 整个短剧技能包的导航文件
-
LICENSE- open-source license
- 开源许可证
-
CHANGELOG.md- version history
- 版本记录
The skills in this repository generally follow these principles:
这个仓库中的 skills 通常遵循以下原则:
If the user gives limited information, generate a strong, usable default version first, then refine only if needed.
如果用户提供的信息有限,先产出一版强可用、能直接工作的默认结果,再视需要继续细化。
Do not silently change the user's request into a different task or genre.
不要擅自把用户原始需求改造成另一个任务、类型或方向。
Add the missing structure needed for execution: goals, constraints, actors, deliverables, quality bars, and next-step logic.
补上执行所需的结构信息:目标、约束、角色、交付物、质量标准以及下一步推进逻辑。
Outputs should be directly reusable in real workflows, not just conceptually better or more verbose.
输出应能直接进入真实工作流,而不只是概念上更好、文字上更长。
A good skill should make strong results repeatable, stable, and easier to extend through references, templates, and examples.
一个好的 skill 应该让高质量结果可复用、可稳定复现,并且可以通过 references、templates 和 examples 持续扩展。
This repository is useful for:
- AI tool builders
- prompt engineers
- agent developers
- content creators
- product managers
- educators
- anyone who writes rough prompts and wants better results
这个仓库适合:
- AI 工具开发者
- Prompt 工程实践者
- Agent 开发者
- 内容创作者
- 产品经理
- 教育工作者
- 任何经常写“粗糙 prompt”但希望结果更好的人
Possible future improvements:
- more examples
- more domain-specific prompt patterns
- more skills in the repository
- richer bilingual prompt templates
- stronger support for coding / design / analysis subtypes
后续可能继续增强:
- 更多 examples
- 更多垂直领域 prompt 模式
- 更多可安装 skills
- 更丰富的双语模板
- 更强的代码 / 设计 / 分析子场景支持
Issues and pull requests are welcome.
欢迎提交 issue 和 pull request。
Good contributions include:
- better bilingual examples
- clearer use cases
- improved metadata
- new reference docs
- additional reusable skills
欢迎的贡献包括:
- 更好的中英双语示例
- 更清晰的 use cases
- 更完善的元信息
- 新的参考文档
- 更多可复用 skills
See:
请查看:
CHANGELOG.md
MIT