QUANTSKILLS 是 AI Agent 时代的开放量化社区，聚焦 Quant Skills（量化技能） 和 Agents（智能体） 两类资产。

QUANTSKILLS 由 PandaAI 发起，连接中文量化开发者与全球 AI 量化社区。PandaAI 在国内通过 PandaAI Quant 服务本土用户，在国际通过 TQX.ai 面向全球开发者与研究者。

我们帮助量化开发者把交易经验、研究方法、因子模型和策略代码，转化为可检索、可安装、可验证、可分享的标准化资产。

把你的量化经验，变成人类可以信任、AI Agent 可以调用的 Skill。

让研究想法，走到真实交付。

QuantStudio 是 QuantSkills / PandaAI 的 AI 量化研究工作台，将技能、专家、专家团、数据与研究产物集中在同一个工作空间。用自然语言描述任务，在对话中推进研究，查看和下载实际生成的报告、图表、代码与文件；支持本地运行，也可部署为团队共享工作区。

查看项目与源码 → · 快速开始

mindmap
  root((QUANTSKILLS))
    🛠️ Skills 技能
      因子计算
      数据清洗
      策略审计
      研报复现
      报告生成
    🤖 Agents 智能体
      研究复现工作流
      策略审计工作流
      内容生成
      社区问答

flowchart LR
    A["💡 创建你的<br/>Skill / Agent"] --> B["🐙 发布到 GitHub"]
    B --> C["📮 提交到<br/>QUANTSKILLS Registry"]
    C --> D["🔍 社区评审与验证"]
    D --> E["🌟 曝光 · 分发<br/>AI Agent 可发现"]

    style A fill:#e3f2fd,stroke:#1976d2
    style E fill:#e8f5e9,stroke:#388e3c

贡献者可以获得：

• 曝光：进入 QUANTSKILLS 目录、官网页面、精选列表与社区推荐
• 可信度：获得 QS-Compatible、PandaData-Compatible、Backtest-Reproducible 等验证标签
• 分发：让 Skill 可被未来的 AI Agent 搜索、安装、调用
• 协作：参与策略共创、内容项目、企业项目、验证服务与付费 Skill
• 个人品牌：从"我写了一个策略"，升级为"我发布了一个被社区收录和评审的量化 Skill"

早期阶段我们不强制统一模板：研究笔记、Prompt、Python 脚本、Agent 工作流、策略代码、数据校验、文档都可以是 Skill。

我们不用模板限制创造力，用注册与验证建立秩序。

QUANTSKILLS 组织下的仓库应使用小写的 skill- 或 agent- 前缀。

• skill-：可复用能力，如因子、策略模板、数据处理、研报复现、验证工具、Prompt、示例或工具。
• agent-：AI Agent 或自动化工作流，如研究复现 Agent、策略审计 Agent、数据处理 Agent、评审 Agent 或多步任务系统。

每个仓库的根目录应包含一个声明文件：

• Skill 仓库：SKILL.md
• Agent 仓库：AGENTS.md

声明文件或项目清单中应包含上游元数据，例如 QuantSkills 组织 URL、仓库名、仓库 URL、项目类型，以及（如适用）所属合集（collection）。

AI 辅助工具可以使用仓库名、SKILL.md / AGENTS.md、README 与描述信息来协助维护公共注册表。最终的收录、推荐、验证或官方认定，仍需经过维护者评审。

完整仓库规则见 COMMUNITY_RULES.md。

flowchart LR
    L1["📋 Level 1 · Listed<br/>基本信息清晰<br/>可进入目录"] --> L2["▶️ Level 2 · Runnable<br/>安装说明 + 示例输入输出<br/>可运行代码 + 依赖信息"]
    L2 --> L3["✅ Level 3 · Verified<br/>数据来源 + 无未来函数检查<br/>回测证据 + 风险说明 + 验证报告"]

    style L1 fill:#f5f5f5,stroke:#9e9e9e
    style L2 fill:#fff3e0,stroke:#f57c00
    style L3 fill:#e8f5e9,stroke:#388e3c

低门槛加入，高标准验证。

QUANTSKILLS 同时为人类和 AI Agent 设计。我们将逐步建设：

• llms.txt
• Skills 索引 / Agents 索引
• MCP 服务
• GitHub README、Topics 与 Release 约定

目标：让 AI Agent 能够从社区搜索、安装、调用、验证量化能力。

公开仓库的元数据、标题、摘要和关键文档以英文为主，方便全球贡献者与 AI Agent 理解和索引；同时支持中文、日文、韩文、西班牙文、法文、德文等语言用于讨论、教程、示例、研究笔记和社区协作。

任何语言的贡献都欢迎，只需附上简短的英文标题、摘要或 README 小节。

• 尊重贡献者，保持建设性讨论。
• 不提交垃圾信息、误导性项目、违法内容、不安全代码、泄露数据或侵权材料。
• 不在公开 Issue、PR、README 或仓库中发布敏感信息（手机号、微信号、邮箱、证件号、密码、API Key、账户凭证）。
• 成员创建的仓库默认为 Community Project，未经评审不得宣称官方、认证、已验证或背书状态。
• 量化项目应明确说明数据来源、假设、局限和风险边界。
• 维护者可在必要时进行内容管理、归档、限制、转移或删除。

完整规则见 COMMUNITY_RULES.md。

成员可在 QUANTSKILLS 组织下创建并维护自己的社区项目。github.com/quantskills 下的仓库由 QUANTSKILLS 组织托管和治理：

• 项目创建者保留作品的署名、荣誉与贡献历史，并可按授予的权限维护仓库；
• 组织所有者保留最终治理权，必要时（安全问题、法律风险、垃圾信息、废弃项目、命名冲突、违反规则）可重命名、归档、转移、限制访问或删除仓库；
• 成员创建的仓库默认为社区项目，不自动代表 QUANTSKILLS 官方验证或背书，后续可按社区规则评审标记为 Listed / Runnable / Verified。

如果你有一个量化方法、因子、策略、工具或工作流，QUANTSKILLS 要帮你把它发布成：人类看得见、AI Agent 找得到、社区可验证的 Skill。

_{扫码加入 PandaAI 社群，交流 QUANTSKILLS 技能、Agent 工作流与量化研究实践。}

QUANTSKILLS is an open community for Quant Skills and Agents in the AI Agent era.

Initiated by PandaAI, QUANTSKILLS connects Chinese quant developers with the global AI quant community. PandaAI serves local users through PandaAI Quant and international developers and researchers through TQX.ai.

We help quant developers turn trading experience, research methods, factor models, and strategy code into standardized assets that can be searched, installed, validated, and shared.

Turn your quant experience into Skills that humans can trust and AI Agents can use.

QUANTSKILLS focuses on two types of assets:

• Skills: factor calculation, data cleaning, strategy audit, research report replication, report generation, and other reusable capability packages
• Agents: research replication, strategy audit, content generation, community Q&A, and other AI Agent workflows

flowchart LR
    A["💡 Create your<br/>Skill / Agent"] --> B["🐙 Publish on GitHub"]
    B --> C["📮 Submit to the<br/>QUANTSKILLS Registry"]
    C --> D["🔍 Community review<br/>& validation"]
    D --> E["🌟 Visibility · Distribution<br/>AI Agent discovery"]

    style A fill:#e3f2fd,stroke:#1976d2
    style E fill:#e8f5e9,stroke:#388e3c

Contributors may gain:

• Visibility: be listed in QUANTSKILLS directories, website pages, curated lists, and community recommendations
• Credibility: earn labels such as QS-Compatible, PandaData-Compatible, Backtest-Reproducible, and other validation marks
• Distribution: make Skills searchable, installable, and callable by future AI Agents
• Collaboration: join strategy co-creation, content projects, enterprise projects, validation services, and paid Skills
• Personal brand: move from "I wrote a strategy" to "I published a quant Skill listed and reviewed by the community"

At the early stage, we do not force every contributor into a single fixed template. Skill formats can be very different: research notes, prompts, Python scripts, agent workflows, strategy code, data checks, or documentation.

We do not use templates to limit creativity. We use registration and validation to build order.

Repositories under the QUANTSKILLS organization should use a lowercase skill- or agent- prefix.

• skill- is for reusable capabilities, such as factors, strategy templates, data processing, report replication, validation utilities, prompts, examples, or tools.
• agent- is for AI Agents or automated workflows, such as research replication agents, strategy audit agents, data processing agents, review agents, or multi-step task systems.

Each repository should include a declaration file at the repository root:

• SKILL.md for Skill repositories
• AGENTS.md for Agent repositories

The declaration file or project manifest should include upstream metadata such as the QuantSkills organization URL, repository name, repository URL, project type, and collection when applicable.

AI-assisted tools may use repository names, SKILL.md / AGENTS.md, README files, and descriptions to help maintain the public registry. Final listing, recommendation, validation, or official recognition still requires maintainer review.

Read the full repository rules: COMMUNITY_RULES.md

flowchart LR
    L1["📋 Level 1 · Listed<br/>clear basic information<br/>listed in the directory"] --> L2["▶️ Level 2 · Runnable<br/>install notes + example I/O<br/>runnable code + dependencies"]
    L2 --> L3["✅ Level 3 · Verified<br/>data sources + no-lookahead checks<br/>backtest evidence + risk notes"]

    style L1 fill:#f5f5f5,stroke:#9e9e9e
    style L2 fill:#fff3e0,stroke:#f57c00
    style L3 fill:#e8f5e9,stroke:#388e3c

Low barrier to join. High standard for validation.

QUANTSKILLS is designed for both humans and AI Agents. We will gradually build:

• llms.txt
• skills index / agents index
• MCP services
• GitHub README, topics, and release conventions

The goal is to let AI Agents search, install, call, and validate quant capabilities from the community.

English is the primary language for public repository metadata, titles, summaries, and key documentation, so global contributors and AI Agents can understand and index the project.

We also support Chinese, Japanese, Korean, Spanish, French, German, and other widely used languages for discussions, tutorials, examples, research notes, and community collaboration.

Contributions in any language are welcome when they include enough English context, such as a short English title, summary, or README section.

• Respect contributors and keep discussions constructive.
• Do not submit spam, misleading projects, illegal content, unsafe code, leaked data, or infringing materials.
• Do not post sensitive information in public Issues, Pull Requests, README files, or repositories.
• Member-created repositories are Community Projects by default and must not claim official, certified, verified, or endorsed status unless reviewed.
• Quant projects should clearly state data sources, assumptions, limitations, and risk boundaries.
• Maintainers may moderate, archive, restrict, transfer, or delete content when necessary.

Read the full rules: COMMUNITY_RULES.md

Members may be allowed to create and maintain their own community projects under the QUANTSKILLS organization.

Repositories created under github.com/quantskills are hosted and governed within the QUANTSKILLS organization. Project creators keep authorship, credit, and contribution history for their work. The project creator may maintain the repository according to the permissions granted to them, while organization owners retain final governance rights.

Member-created repositories are Community Projects by default. They do not automatically represent official QUANTSKILLS validation or endorsement. Projects may later be reviewed and marked as Listed, Runnable, or Verified according to community rules.

Organization owners may rename, archive, transfer, restrict access to, or delete repositories when necessary, especially for security issues, legal risk, spam, abandoned projects, naming conflicts, or violations of community rules.

If you have a quant method, factor, strategy, tool, or workflow, QUANTSKILLS should help you publish it as a Skill that humans can see, AI Agents can discover, and the community can validate.

_{Scan the QR code to join the PandaAI community for QUANTSKILLS skills, agent workflows, and quantitative research practice.}