WorldQuant BRAIN Skills

基于 Claude Skills 的量化因子自动挖掘框架

集成 WorldQuant BRAIN (cnhkmcp) + Claude Code/Gemini CLI/iFlow，实现因子研究全流程自动化

English | 中文

---

中文

📚 目录

• 快速开始
• 核心 Skills 介绍
  • Alpha Research Recorder
  • Factor Backtest
  • Knowledge Base Search

---

快速开始

5 分钟快速上手指南：

1. 安装与配置

# 克隆仓库
cd worldquant-skill

# 安装 WorldQuant BRAIN MCP 工具
pip install cnhkmcp

# 复制 skills 到对应目录
# - Claude Code: 复制到 ~/.claude/skills/
# - iFlow: 复制到 ~/.iflow/skills/
# - Gemini CLI: 复制到 ~/.gemini/skills/

2. 核心 Skills 快速体验

方式 1：交互式使用（Claude Code / Gemini CLI / iFlow）

# 在 Claude Code 中
"使用 alpha-research-recorder skill，帮我记录这次研究会话"
"使用 factor_backtest skill，帮我回测这 8 个表达式"
"使用 knowledge_base_search skill，搜索降低 turnover 的方法"

# 在 Gemini CLI / iFlow 中（类似）

方式 2：工作流集成（通过 .md 文件定义 SOP）

# factor-mining-workflow.md

## 目标
自动化因子挖掘流程

## 步骤

### 1. 搜索优化方法
使用 skill: knowledge_base_search
搜索降低 turnover 的方法

### 2. 记录研究会话
使用 skill: alpha-research-recorder
记录研究配置和约束

### 3. 批量回测
使用 skill: factor_backtest
回测 8 个 Alpha 表达式

## 在命令行工具中执行
# Claude Code / Gemini CLI / iFlow:
"执行 factor-mining-workflow.md 中定义的流程"

---

核心 Skills 介绍

本项目包含三个核心 Skills，专门为 WorldQuant BRAIN 平台的量化因子研究设计，支持两种使用方式：

• 交互式使用：在 Claude Code、Gemini CLI、iFlow 等命令行工具中直接对话调用
• 工作流集成：在命令行工具中通过 .md 文件定义 SOP（标准操作流程），让 Agent 按流程自动完成任务

1. Alpha Research Recorder

一句话总结：自动记录和保存您的 Alpha 研究过程，让每一步都有迹可循。

核心功能：

• 根据记录类型（session_meta / round / final_summary）引用对应模板文件
• 接收结构化数据并验证必填字段
• 创建/更新 YAML 或 Markdown 日志文件

使用场景：

• 研究开始时记录配置和约束（session_meta）
• 单轮研究结束后记录完整过程（round）
• 研究完成后生成总结报告（final_summary）

文件结构：

logs/
└── 20260108_analyst_sonnet/
    ├── session_metadata.yml    # 会话配置
    ├── round_0001.yml          # 第1轮记录
    ├── round_0002.yml          # 第2轮记录
    └── final_summary.md        # 最终总结

使用方式：

"使用 alpha-research-recorder skill，帮我记录这次研究会话"

详细文档：alpha-research-recorder/SKILL.md

---

2. Factor Backtest

一句话总结：批量回测 8 个 Alpha 表达式，自动验证并监控进度。

核心功能：

• 验证表达式的合法性
• 提交批量回测任务
• 监控进度并处理异常
• 返回完整的回测结果

核心约束：

• ⚠️ Rule of 8：必须恰好传入 8 个表达式
• ⚠️ 加速配置：必须设置 visualization=false
• ⚠️ 中性化选择：必须从 5 个选项中选择

标准 4 步流程：

1. 验证表达式（validate_alpha_expressions）
2. 提交回测（create_multiSim）
3. 监控进度（check_multisimulation_status）
4. 获取结果（get_multisimulation_result）

使用方式：

"使用 factor_backtest skill，帮我回测这 8 个表达式：[表达式列表]"

详细文档：factor_backtest/SKILL.md

---

3. Knowledge Base Search

一句话总结：智能搜索知识库，快速找到数据字段、优化方法和优质示例。

核心功能：

• 支持关键词搜索、语义理解、字段查找和示例推荐
• 涵盖数据集字段、优化经验、优质示例和平台机制
• 返回结构化、可追溯的搜索结果

知识库结构：

./Resources/
├── DATASET/                    # 数据集目录
├── good_alpha_examples/             # 优质示例
├── How_WorldQuant_BRAIN_Backtesting_Works.md  # 平台机制
├── alpha_optimization/              # 优化经验
└── regular_operators.csv           # 操作符清单

使用方式：

"使用 knowledge_base_search skill，搜索降低 turnover 的方法"
"使用 knowledge_base_search skill，查找 close 字段的信息"

详细文档：knowledge_base_search/SKILL.md

---

English

A Quantitative Factor Auto-Mining Framework Based on Claude Skills

Integrating WorldQuant BRAIN (cnhkmcp) + Claude Code/Gemini CLI/iFlow for automated factor research workflow

📚 Table of Contents

• Quick Start
• Core Skills Overview
  • Alpha Research Recorder
  • Factor Backtest
  • Knowledge Base Search

---

Quick Start

5-minute quick start guide:

1. Installation & Configuration

# Clone repository
git clone https://github.com/your-org/worldquant-skill.git
cd worldquant-skill

# Install WorldQuant BRAIN MCP tool
pip install cnhkmcp

# Copy skills to corresponding directories
# - Claude Code: copy to ~/.claude/skills/
# - iFlow: copy to ~/.iflow/skills/
# - Gemini CLI: copy to ~/.gemini/skills/

2. Core Skills Quick Experience

Method 1: Interactive Usage (Claude Code / Gemini CLI / iFlow)

# In Claude Code
"Use alpha-research-recorder skill to record this research session"
"Use factor_backtest skill to backtest these 8 expressions"
"Use knowledge_base_search skill to search for turnover reduction methods"

# In Gemini CLI / iFlow (similar)

Method 2: Workflow Integration (Define SOP via .md files)

# factor-mining-workflow.md

## Goal
Automated factor mining workflow

## Steps

### 1. Search optimization methods
Use skill: knowledge_base_search
Search for turnover reduction methods

### 2. Record research session
Use skill: alpha-research-recorder
Record research configuration and constraints

### 3. Batch backtest
Use skill: factor_backtest
Backtest 8 Alpha expressions

## Execute in command-line tools
# Claude Code / Gemini CLI / iFlow:
"Execute the workflow defined in factor-mining-workflow.md"

---

Core Skills Overview

This project includes three core Skills designed specifically for quantitative factor research on the WorldQuant BRAIN platform, supporting two usage methods:

• Interactive Usage: Direct conversational invocation in Claude Code, Gemini CLI, iFlow, and other command-line tools
• Workflow Integration: Define SOPs (Standard Operating Procedures) through .md files in command-line tools, enabling Agents to complete tasks automatically according to the workflow

1. Alpha Research Recorder

One-line Summary: Automatically record and save your Alpha research process, making every step traceable.

Core Features:

• Reference corresponding template files based on record type (session_meta / round / final_summary)
• Receive structured data and validate required fields
• Create/update YAML or Markdown log files

Use Cases:

• Record configuration and constraints at research start (session_meta)
• Record complete process after single research round (round)
• Generate summary report after research completion (final_summary)

File Structure:

logs/
└── 20260108_analyst_sonnet/
    ├── session_metadata.yml    # Session configuration
    ├── round_0001.yml          # Round 1 record
    ├── round_0002.yml          # Round 2 record
    └── final_summary.md        # Final summary

Usage:

"Use alpha-research-recorder skill to record this research session"

Detailed Documentation: alpha-research-recorder/SKILL.md

---

2. Factor Backtest

One-line Summary: Batch backtest 8 Alpha expressions with automatic validation and progress monitoring.

Core Features:

• Validate expression legality
• Submit batch backtest tasks
• Monitor progress and handle exceptions
• Return complete backtest results

Core Constraints:

• ⚠️ Rule of 8: Must pass exactly 8 expressions
• ⚠️ Acceleration Config: Must set visualization=false
• ⚠️ Neutralization Selection: Must choose from 5 options

Standard 4-Step Workflow:

1. Validate expressions (validate_alpha_expressions)
2. Submit backtest (create_multiSim)
3. Monitor progress (check_multisimulation_status)
4. Get results (get_multisimulation_result)

Usage:

"Use factor_backtest skill to backtest these 8 expressions: [expression list]"

Detailed Documentation: factor_backtest/SKILL.md

---

3. Knowledge Base Search

One-line Summary: Intelligently search the knowledge base to quickly find data fields, optimization methods, and quality examples.

Core Features:

• Support keyword search, semantic understanding, field lookup, and example recommendations
• Cover dataset fields, optimization experience, quality examples, and platform mechanisms
• Return structured, traceable search results

Knowledge Base Structure:

./Resources/
├── DATASET/                    # Dataset directory
├── good_alpha_examples/             # Quality examples
├── How_WorldQuant_BRAIN_Backtesting_Works.md  # Platform mechanisms
├── alpha_optimization/              # Optimization experience
└── regular_operators.csv           # Operator list

Usage:

"Use knowledge_base_search skill to search for turnover reduction methods"
"Use knowledge_base_search skill to find information about close field"

Detailed Documentation: knowledge_base_search/SKILL.md

---

Contributing

Contributions welcome! Please follow these steps:

1. Fork this repository
2. Create feature branch (git checkout -b feature/amazing-skill)
3. Commit changes (git commit -m 'feat: add amazing skill')
4. Push to branch (git push origin feature/amazing-skill)
5. Create Pull Request

License

This project is licensed under the Apache 2.0 License. See LICENSE file for details.

---

⬆ Back to Top

Made with ❤️ by the WorldQuant BRAIN community