Give AI experts the thinking quality of historical giants — Enhance task execution quality through name-based quality anchors
让 AI 专家拥有历史巨匠的思维品质 — 通过人名质量锚点提升任务执行质量
Give AI experts the thinking quality of historical giants
让 AI 专家拥有历史巨匠的思维品质
The project author has no programming or coding knowledge whatsoever. This is purely product-thinking driven.
本项目作者完全不懂程序和代码，纯属产品思维驱动。
- Lead Development: Qwen 3.5 Plus
主导开发: Qwen 3.5 Plus - Strategic Architecture: Qwen-Max
战略架构: Qwen-Max - Deep Research: DeepSeek 深度研究: DeepSeek
- Innovation Suggestions: Gemini 创新建议: Gemini
This is a truly collaborative open-source project built entirely by an AI expert team!
这是一个真正由 AI 专家团队协作完成的开源项目！
The effectiveness of names as "quality anchors" is based on four major scientific research findings:
人名作为"质量锚点"的有效性基于四大科学研究发现：
"Detailed expert descriptions significantly improve LLM answer quality, with ExpertLLaMA reaching 96% of ChatGPT's capability"
来源: arXiv:2305.14688
"详细专家描述可显著提升 LLM 回答质量，ExpertLLaMA 达到 ChatGPT 96% 能力"
来源: arXiv:2305.14688
Why it works: Names serve as expert identity shortcuts, activating relevant knowledge and thinking patterns stored in the LLM.
为什么有效: 人名作为专家身份的快捷标识，激活 LLM 中存储的相关知识和思维模式。
"Role-playing prompts significantly improve performance on reasoning tasks: AQuA +10.3%, Last Letter +60.4%"
来源: arXiv:2308.07702
"角色扮演提示在推理任务上显著提升性能：AQuA +10.3%，Last Letter +60.4%"
来源: arXiv:2308.07702
Why it works: Names provide specific thinking frameworks and behavioral patterns, more effectively triggering Chain-of-Thought reasoning than simple "think step by step" prompts.
为什么有效: 人名提供具体的思维框架和行为模式，比简单的"think step by step"更有效地触发 Chain-of-Thought 推理过程。
"Personas can improve or degrade performance, with degradation observed in 7 out of 12 datasets"
来源: arXiv:2408.08631
"Persona 可能提升也可能降低表现，在 7/12 数据集中表现下降"
来源: arXiv:2408.08631
Our solution: Using historically validated successful figures as quality anchors avoids the risks of arbitrary personas.
我们的解决方案: 使用历史验证的成功人物作为质量锚点，避免随意 persona 带来的风险。
"Role prompts are effective for style but have limited impact on accuracy"
来源: arXiv:2406.06608
"角色提示对风格有效，对准确性效果有限"
来源: arXiv:2406.06608
Our innovation: Emphasizing quality anchors rather than role-playing, focusing on "deliver iPhone-level results" rather than "speak like Jobs".
我们的创新: 强调质量锚点而非角色扮演，关注"产出 iPhone 级别的结果"而非"像乔布斯一样说话"。
| Aspect | Traditional Role-Playing | Our Quality Anchors |
|---|---|---|
| Goal | Simulate role behavior | Achieve quality standards |
| Risk | May activate negative stereotypes | Based on historical success cases |
| Stability | Unstable effects | Historically validated reliability |
| Task Focus | Focus on "how to speak" | Focus on "what to do" |
| 方面 | 传统角色扮演 | 我们的质量锚点 |
|---|---|---|
| 目标 | 模拟角色行为 | 达到质量标准 |
| 风险 | 可能激活负面刻板印象 | 基于历史成功案例 |
| 稳定性 | 效果不稳定 | 历史验证可靠 |
| 任务导向 | 关注"如何说" | 关注"做什么" |
Key Insight: Names don't make AI "impersonate" someone, but rather "achieve" the historical achievement standards created by that person.
关键洞察: 人名不是让 AI "扮演"某个人，而是让 AI "达到"某个人创造的历史成就标准。
Current Research Status:
当前研究状态:
While there are currently no academic papers directly studying the combined effect of "expert role cards + names," the following research provides our theoretical foundation:
虽然目前没有直接研究"专家角色卡 + 人名"组合效果的学术论文，但以下研究为我们提供了理论基础：
-
Jekyll & Hyde Framework (Kim et al., 2024): Demonstrates that combining persona + neutral prompts is more effective than single strategies, improving robustness by generating multiple results simultaneously and selecting the optimal one [arXiv:2408.08631]
-
Jekyll & Hyde 框架 (Kim et al., 2024): 证明组合 persona + 中性提示比单一策略更有效，通过同时生成多种结果并选择最优解来提升鲁棒性 [arXiv:2408.08631]
-
Multi-Agent Collaboration (Wang et al., 2024): Shows that multi-persona self-collaboration can unleash LLMs' cognitive synergy effects [arXiv:2307.05300]
-
多智能体协作 (Wang et al., 2024): 证明多个人格自我协作可以释放 LLM 的认知协同效应 [arXiv:2307.05300]
-
Ensemble Methods: Proves that aggregating different strategies can reduce variance, enhance robustness, and capture complementary strengths
-
集成方法: 证明不同策略的聚合可以减少方差、增强鲁棒性、捕获互补优势
Our Hypothesis:
我们的假设:
Expert Role Cards (domain knowledge framework) + Name Quality Anchors (historical success cases) = Synergistic Effect (1+1>2)
专家角色卡（领域专业知识框架）+ 人名质量锚点（历史成功案例）= 协同效应（1+1>2）
Honest Declaration:
诚实声明:
This project was completed in 5 hours and some parts are still rough. I cannot conduct quantitative comparison experiments. I sincerely hope researchers capable of conducting controlled experiments will participate in the project, validate the combination effects through A/B testing, and help improve project quality. Pull Requests and Issues are welcome!
本项目在5小时内完成，部分地方较为粗糙。我无法进行量化对比实验，诚挚希望有能力进行对照实验的研究者参与项目，通过 A/B 测试验证组合效果，并帮助完善项目质量。欢迎提交 Issue 或 Pull Request！
- Author: Alireza Rezvani (CTO and AI engineer based in Berlin)
作者: Alireza Rezvani (柏林 CTO 和 AI 工程师) - Stars: 6,000+ GitHub stars ⭐
Stars: 6,000+ GitHub stars ⭐ - URL: https://github.com/alirezarezvani/claude-skills
URL: https://github.com/alirezarezvani/claude-skills - Contribution: Provides 205 production-ready professional skill packages covering 9 major domains
贡献: 提供 205 个生产就绪的专业技能包，覆盖 9 大领域
Our Innovation:
我们的创新:
Building upon Claude Skills' expert role cards, we introduced the "name quality anchor" concept, transforming abstract professional capabilities into concrete historical success cases.
在 Claude Skills 的专家角色卡基础上，我们引入了"人名质量锚点"概念，将抽象的专业能力转化为具体的历史成功案例。
# Clone this repository
# 克隆本仓库
git clone https://github.com/your-username/expert-library-plus.git
# Copy expert library to OpenClaw directory
# 复制专家库到 OpenClaw 目录
cp -r expert-library-plus/experts/ ~/.openclaw/experts/
# Verify installation (should see expert list)
# 验证安装（应该看到专家列表）
ls ~/.openclaw/experts/Please refer to Windows Installation Guide for detailed Windows installation steps.
请参考 Windows 安装指南 获取详细的 Windows 安装步骤。
💡 User-Friendly Note: OpenClaw currently requires manual file copying to the experts directory. We're exploring automatic discovery mechanisms, but this is currently the most reliable approach.
💡 用户友好提示: OpenClaw 目前需要手动复制文件到专家目录。我们正在探索自动发现机制，但目前这是最可靠的方式。
Please help me design a revolutionary product with expert assistance
请专家帮我设计一个革命性的产品
The system will intelligently recommend the most suitable quality anchors based on your task type:
系统会根据你的任务类型，智能推荐最适合的质量锚点：
Mode A: Name-Only Mode (Lightweight)
模式 A: 仅人名模式（轻量级）
- Use names directly as quality anchors
直接使用人名作为质量锚点 - Ideal for simple tasks or quick validation
适合简单任务或快速验证 - Example:
"Help me design a product from Jobs' perspective"
示例："用乔布斯视角帮我设计产品"
Mode B: Full Loading Mode (Professional)
模式 B: 完整加载模式（专业级）
- Load complete expert role cards + name quality anchors
加载完整专家角色卡 + 人名质量锚点 - Includes conflict detection and trait matching
包含冲突检测和特质匹配 - Ideal for complex tasks or high-quality requirements
适合复杂任务或高质量要求
The system automatically analyzes:
系统会自动分析：
- Personality Trait Conflicts: Avoids contradictory expert combinations
人物特质冲突: 避免选择相互矛盾的专家组合 - Task Match Score: Recommends names most suitable for current tasks
任务匹配度: 推荐最适合当前任务的人名 - Domain Relevance: Ensures names are highly relevant to expert domains
领域适配性: 确保人名与专家领域高度相关
For other users: You can absolutely create your own name library!
对于其他用户：你完全可以创建自己的人名库！
- Directory Structure: Create
names/folders under corresponding expert directories
目录结构: 在对应专家目录下创建names/文件夹 - File Format: Reference
templates/name-template.md
文件格式: 参考templates/name-template.md - Automatic Integration: The system automatically discovers and loads your name libraries
自动关联: 系统会自动发现并加载你的人名库 - No Code Changes: No core code modification needed, fully plugin-based
无需修改: 无需修改任何核心代码，完全插件化
Example: If you want to add new names for frontend development experts, simply create new
.mdfiles in theengineering/names/directory, and the system will automatically recognize and provide them as options!
示例: 如果你想为前端开发专家添加新的人名，只需在engineering/names/目录下创建新的.md文件，系统会自动识别并提供选择！
expert-library-plus/
├── experts/ # 43+ professional experts
│ ├── engineering/ # Engineering experts (frontend, backend, security, etc.)
│ │ └── names/ # Name library (user-extensible)
│ ├── design/ # Design experts (UI/UX, branding, etc.)
│ │ └── names/ # Name library (user-extensible)
│ ├── business/ # Business experts (sales, marketing, strategy, etc.)
│ │ └── names/ # Name library (user-extensible)
│ └── safety/ # Safety experts (security engineering, incident response, etc.)
│ └── names/ # Name library (user-extensible)
├── templates/ # Create your own experts
│ └── name-template.md # Detailed creation guide
└── [root files] # README, LICENSE, CONTRIBUTING, etc.
expert-library-plus/
├── experts/ # 43+ 个专业专家
│ ├── engineering/ # 工程专家（前端、后端、安全等）
│ │ └── names/ # 人名库（用户可扩展）
│ ├── design/ # 设计专家（UI/UX、品牌等）
│ │ └── names/ # 人名库（用户可扩展）
│ ├── business/ # 商业专家（销售、营销、策略等）
│ │ └── names/ # 人名库（用户可扩展）
│ └── safety/ # 安全专家（安全工程、事件响应等）
│ └── names/ # 人名库（用户可扩展）
├── templates/ # 创建你自己的专家
│ └── name-template.md # 详细创建指南
└── [根目录文件] # README, LICENSE, CONTRIBUTING 等
- Auto-Discovery: System automatically scans all
names/directories
自动发现: 系统自动扫描所有names/目录 - Zero Configuration: No code changes needed, just add files
零配置: 无需代码更改，只需添加文件 - Hot Loading: New name files are immediately available
热加载: 新人名文件立即可用 - Complete Isolation: User-created content is completely separated from core system
完全隔离: 用户创建内容与核心系统完全分离
- Deceased celebrities only: Demo files contain only deceased historical figures
仅用已故名人: 示范文件只包含已故的历史人物 - User self-building mechanism: You have complete control over which names to add
用户自建机制: 你完全控制添加哪些人名 - Legal disclaimer: Clear statement of fair use principles
法律免责声明: 明确说明合理使用原则
- MIT License: Allows commercial use and modification
MIT 许可证: 允许商业使用和修改 - Global community: Bilingual Chinese-English support, welcoming global contributors
全球社区: 中英文双语支持，欢迎全球贡献者
This project is built upon Claude Code Skills & Plugins — Agent Skills for Every Coding Tool (6,000+ GitHub stars), with special thanks to Alireza Rezvani for creating this outstanding open-source skill library.
本项目基于 Claude Code Skills & Plugins — Agent Skills for Every Coding Tool（6,000+ GitHub stars）的坚实基础，特别感谢 Alireza Rezvani 创建了这个卓越的开源技能库。
Claude Skills Contributions:
Claude Skills 的贡献:
- Provides 205 production-ready professional skill packages
提供了 205 个生产就绪的专业技能包 - Covers 9 major domains: engineering, product, marketing, compliance, C-level advisory, etc.
覆盖工程、产品、营销、合规、高管顾问等 9 大领域 - Supports 11 AI coding tools (Claude Code, Codex, Gemini CLI, etc.)
支持 11 种 AI 编码工具（Claude Code、Codex、Gemini CLI 等）
Our Innovation:
我们的创新:
Based on Claude Skills' expert role cards, we innovatively introduced the "name as quality anchor" concept, transforming abstract professional capabilities into concrete historical success cases, enabling AI experts to achieve the thinking quality and work standards of historical giants.
基于 Claude Skills 的专家角色卡，我们创新性地引入了"人名作为质量锚点"概念，将抽象的专业能力转化为具体的历史成功案例，让 AI 专家能够达到历史上巨匠的思维品质和工作标准。
For detailed usage examples, troubleshooting, and expected effects, please refer to Usage Examples Guide.
详细使用示例、故障排除和预期效果请参考 使用示例指南.
For detailed version compatibility and system requirements, please refer to Compatibility Guide.
详细版本兼容性和系统要求请参考 兼容性指南.
Don't just read—try it immediately!
不要只是阅读——立即尝试！
- Clone this repository
克隆本仓库 - Copy to your OpenClaw experts directory
复制到你的 OpenClaw 专家目录 - Say to AI: "Please help me with expert assistance..."
对 AI 说："请专家帮我..."
You'll immediately experience a significant improvement in AI expert thinking quality!
你将立即体验到 AI 专家思维品质的显著提升！
We believe: The best projects come from real user feedback.
我们相信：最好的项目来自于真实的用户反馈。
- Having issues? Submit an Issue!
遇到问题？提交 Issue！ - Have new ideas? Submit a Pull Request!
有新想法？提交 Pull Request！ - Want new experts? Let us know!
想要新专家？告诉我们！
Let's build the world's most powerful AI expert library together!
让我们一起打造世界上最强大的 AI 专家库！