quant_trading-skills

获取股票/基金等金融标的的量化数据，包括行情数据、财务数据、资金流向数据和舆情数据。支持单只股票查询和批量数据拉取。当用户需要查询股票行情、财务指标、资金流向或相关舆情信息时使用此技能。

Security Scan

The skill's code largely matches its stated purpose (fetching market/financial/sentiment data) but there are metadata/instruction mismatches and operational behaviours (local file writes, running bundled Python, installing PyPI packages, child process execution, logging of full context) that the publisher did not declare — review before installing or running.

Purpose & Capability

The code implements exactly the described functionality (market/finance/fund_flow/public_opinion and batch pulls using akshare) and stores results locally in Parquet format. However the registry metadata declared no required binaries or config paths while the code requires Python >=3.11, Node.js and writes to config/ and data/ paths — the runtime requirements are not reflected in the declared metadata.

Instruction Scope

SKILL.md instructs the agent to call the exported execute function and documents parameters and outputs. The runtime instructions do not ask the agent to read unrelated system files or external secrets. The code does log the full context and parameters (index.js calls log with { params, context }), which could include sensitive runtime context if present — SKILL.md/metadata do not call this out.

Install Mechanism

There is no install spec in the registry, but the bundle contains install.py, requirements.txt, package.json and README install instructions. install.py will upgrade pip and install packages from PyPI (akshare, pandas, pyarrow). That is a normal mechanism for this skill's Python components but is not declared in the registry metadata. Installing will fetch third‑party packages from PyPI (network required).

Credentials

The skill does not request environment variables or credentials (none declared). It does, however, read and write local files (config/fetch_status.json, config/data_config.json, and various data/ directories). The code also logs the incoming context object — depending on platform, 'context' can contain sensitive information; the skill will include that in logs. No unrelated external credentials are requested.

Persistence & Privilege

always:false and user-invocable:true — reasonable. The skill writes local files (data/, config/) and may run long‑running batch jobs; it does not modify other skills or global agent configuration. Running large batch jobs may consume significant CPU, time and network bandwidth.

What to consider before installing

This skill appears to implement the advertised quantitative-data functionality, but the package metadata omits important operational details. Before installing or running: 1) Review lib/utils.js (how logging is implemented) to ensure logs are not sent to remote endpoints and that sensitive 'context' is not leaked. 2) Expect to run Python (requires >=3.11) and Node.js; the included install.py will upgrade pip and install packages from PyPI (akshare, pandas, pyarrow) — run in a controlled/sandboxed environment. 3) The skill writes data to local folders (data/, config/); ensure you are comfortable with those files being created and inspect their contents/permissions. 4) Batch operations iterate over all A‑share symbols and can be long-running and network-intensive — run with limits or on a machine with appropriate resources. 5) If you need higher assurance, ask the publisher to: declare required binaries/config paths in metadata, remove or explain any telemetry in logging, and provide an install manifest (or container) rather than relying on a pip upgrade/install script. Running the code in an isolated environment (container/VM) and doing a quick code review of lib/utils.js and any logging/print paths will materially reduce risk.

量化数据获取

概述

本 Skill 用于获取股票/基金等金融标的的量化数据，支持以下四类数据：

1. 行情数据 (market)：获取股票历史行情数据，包括开盘价、收盘价、最高价、最低价、成交量等
2. 财务数据 (finance)：获取股票财务指标数据，包括营业收入、净利润、ROE、资产负债率等
3. 资金流向数据 (fund_flow)：获取个股资金流向数据，包括主力净流入、超大单净流入等
4. 舆情数据 (public_opinion)：获取相关舆情信息

使用说明

基本调用方式

OpenClaw 通过调用 index.js 中导出的 execute 函数来使用此 Skill：

const { execute } = require('./quant_data_fetch');

const result = await execute(params, context);

输入参数

参数名	类型	必填	说明	可选值
type	string	是	数据类型	market, finance, fund_flow, public_opinion
symbol	string	是	标的代码	例如 600519（贵州茅台）
period	string	否	行情周期（仅 market 类型需要）	1d, 1w, 1M
report_type	string	否	报告类型（仅 finance 类型需要）	quarter, year
start_date	string	否	开始日期	格式：YYYY-MM-DD
end_date	string	否	结束日期	格式：YYYY-MM-DD

输出格式

{
  "success": true,
  "data": {...},
  "message": "数据获取成功"
}

数据类型详解

1. 行情数据 (market)

获取股票历史行情数据。

调用示例：

const result = await execute({
  type: 'market',
  symbol: '600519',
  period: '1d',
  start_date: '2024-01-01',
  end_date: '2024-01-31'
});

返回数据示例：

{
  "success": true,
  "data": {
    "symbol": "600519",
    "period": "1d",
    "items": [
      {
        "date": "2024-01-01",
        "open": 1700.0,
        "high": 1750.0,
        "low": 1690.0,
        "close": 1730.0,
        "volume": 1000000,
        "amount": 1730000000,
        "change": 30.0,
        "change_percent": 1.76,
        "turnover": 0.5
      }
    ]
  },
  "message": "数据获取成功"
}

字段说明：

• date: 日期
• open: 开盘价
• high: 最高价
• low: 最低价
• close: 收盘价
• volume: 成交量
• amount: 成交额
• change: 涨跌额
• change_percent: 涨跌幅(%)
• turnover: 换手率(%)

2. 财务数据 (finance)

获取股票财务指标数据。

调用示例：

const result = await execute({
  type: 'finance',
  symbol: '600519',
  report_type: 'quarter'
});

返回数据示例：

{
  "success": true,
  "data": {
    "symbol": "600519",
    "report_type": "quarter",
    "items": [
      {
        "report_date": "2023-12-31",
        "revenue": 120000000000,
        "gross_profit": 80000000000,
        "net_profit": 40000000000,
        "eps": 32.0,
        "roe": 25.0,
        "asset_liability_ratio": 30.0,
        "operating_cash_flow": 35000000000
      }
    ]
  },
  "message": "数据获取成功"
}

字段说明：

• report_date: 报告日期
• revenue: 营业收入(元)
• gross_profit: 营业利润(元)
• net_profit: 净利润(元)
• eps: 每股收益(元)
• roe: 净资产收益率(%)
• asset_liability_ratio: 资产负债率(%)
• operating_cash_flow: 经营活动产生的现金流量净额(元)

3. 资金流向数据 (fund_flow)

获取个股资金流向数据。

调用示例：

const result = await execute({
  type: 'fund_flow',
  symbol: '600519'
});

返回数据示例：

{
  "success": true,
  "data": {
    "symbol": "600519",
    "items": [
      {
        "date": "2024-01-01",
        "main_inflow": 50000000,
        "main_inflow_rate": 2.5,
        "超大_inflow": 30000000,
        "超大_inflow_rate": 1.5,
        "large_inflow": 20000000,
        "large_inflow_rate": 1.0,
        "medium_inflow": -5000000,
        "medium_inflow_rate": -0.25,
        "small_inflow": -45000000,
        "small_inflow_rate": -2.25
      }
    ]
  },
  "message": "数据获取成功"
}

字段说明：

• date: 日期
• main_inflow: 主力净流入-净额(万元)
• main_inflow_rate: 主力净流入-净占比(%)
• 超大_inflow: 超大单净流入-净额(万元)
• 超大_inflow_rate: 超大单净流入-净占比(%)
• large_inflow: 大单净流入-净额(万元)
• large_inflow_rate: 大单净流入-净占比(%)
• medium_inflow: 中单净流入-净额(万元)
• medium_inflow_rate: 中单净流入-净占比(%)
• small_inflow: 小单净流入-净额(万元)
• small_inflow_rate: 小单净流入-净占比(%)

4. 舆情数据 (public_opinion)

获取相关舆情信息。

调用示例：

const result = await execute({
  type: 'public_opinion',
  symbol: '600519'
});

返回数据示例：

{
  "success": true,
  "data": {
    "symbol": "600519",
    "items": [
      {
        "date": "2024-01-01",
        "source": "东方财富网",
        "title": "贵州茅台发布2024年第一季度财报，净利润同比增长20%",
        "sentiment": "positive",
        "url": "https://www.eastmoney.com/600519"
      }
    ]
  },
  "message": "数据获取成功"
}

字段说明：

• date: 日期
• source: 来源
• title: 标题
• sentiment: 情感倾向 (positive/negative/neutral)
• url: 链接

安装要求

• Node.js: >= 16.0.0
• Python: >= 3.11.0

安装步骤

1. 安装 Node.js 依赖

   npm install
   

2. 安装 Python 依赖

   python install.py
   

   或

   pip install -r requirements.txt
   

目录结构

quant_trading-skills/
├── SKILL.md              # 技能核心定义文档
├── README.md             # 使用说明
├── package.json          # Node.js 依赖配置
├── requirements.txt      # Python 依赖配置
├── install.py            # 自动安装脚本
├── index.js              # 主入口文件
├── lib/                  # 核心库
│   ├── utils.js          # 工具函数
│   └── dataFetcher.js    # 数据拉取函数
└── scripts/              # 执行脚本
    └── akshare_fetcher.py # Python 数据拉取脚本

缓存机制

本 Skill 实现了内存缓存机制，相同参数的请求会缓存 1 小时，避免重复请求相同数据。

异常处理

Skill 会捕获并处理所有异常，返回统一的错误格式：

{
  "success": false,
  "data": null,
  "message": "执行失败: 错误信息"
}

批量拉取功能

本 Skill 支持批量拉取各类数据，适用于数据初始化或增量更新场景。

批量拉取类型

类型	说明	数据源
batch_market	批量拉取A股行情数据	akshare
batch_north_flow	批量拉取北向资金数据	akshare
batch_lhb	批量拉取龙虎榜数据	akshare
batch_sentiment	批量拉取舆情数据	akshare
batch_financial	批量拉取财报数据	akshare
stock_codes	获取全部A股股票代码列表	akshare

批量拉取调用示例

1. 批量拉取行情数据

const result = await execute({
  type: 'batch_market',
  batch_market: {
    default_years: 5,
    max_retries: 3
  }
});

返回数据示例：

{
  "success": true,
  "data": {
    "total_stocks": 5000,
    "success_count": 4980,
    "failed_count": 20,
    "total_records": 5000000,
    "is_incremental": false,
    "elapsed_seconds": 1800.5
  },
  "message": "批量拉取完成，成功 4980 只，失败 20 只"
}

2. 批量拉取北向资金数据

const result = await execute({
  type: 'batch_north_flow',
  batch_north_flow: {
    default_years: 5
  }
});

3. 批量拉取龙虎榜数据

const result = await execute({
  type: 'batch_lhb',
  batch_lhb: {
    default_years: 3
  }
});

4. 批量拉取舆情数据

const result = await execute({
  type: 'batch_sentiment',
  batch_sentiment: {
    default_years: 2
  }
});

5. 批量拉取财报数据

const result = await execute({
  type: 'batch_financial',
  batch_financial: {
    default_years: 3,
    max_retries: 3
  }
});

6. 获取股票代码列表

const result = await execute({
  type: 'stock_codes'
});

返回数据示例：

{
  "success": true,
  "data": {
    "items": [
      { "code": "000001", "name": "平安银行" },
      { "code": "000002", "name": "万科A" }
    ],
    "total": 5000
  },
  "message": "数据获取成功"
}

增量拉取机制

本 Skill 实现了智能增量拉取机制，避免重复拉取已有数据。

工作原理

1. 状态记录：每次批量拉取完成后，会在 config/fetch_status.json 中记录拉取时间
2. 日期计算：下次拉取时，根据上次拉取时间计算需要获取的日期范围
3. 增量标识：返回结果中 is_incremental 字段标识是否为增量拉取

状态文件结构

{
  "market": {
    "last_fetch_time": "2024-01-15 10:30:00",
    "fetch_count": 5,
    "last_fetch_count": 4980
  },
  "north_flow": {
    "last_fetch_time": "2024-01-15 11:00:00",
    "fetch_count": 3,
    "last_fetch_count": 1200
  },
  "lhb": {
    "last_fetch_time": "2024-01-15 11:30:00",
    "fetch_count": 2,
    "last_fetch_count": 500
  },
  "sentiment": {
    "last_fetch_time": "2024-01-15 12:00:00",
    "fetch_count": 1,
    "last_fetch_count": 10000
  },
  "financial": {
    "last_fetch_time": "2024-01-15 12:30:00",
    "fetch_count": 1,
    "last_fetch_count": 5000
  }
}

增量拉取逻辑

• 首次拉取：is_incremental: false，拉取 default_years 年的数据
• 增量拉取：is_incremental: true，仅拉取上次拉取时间之后的新数据
• 时间间隔：从上次拉取时间的下一天开始，到当前日期结束

Parquet 存储格式

所有批量拉取的数据均采用 Apache Parquet 格式存储，具有以下优势：

存储优势

• 高效压缩：使用 Snappy 压缩算法，压缩率高
• 列式存储：支持列级别的数据读取，查询效率高
• 跨平台兼容：支持 Python、Java、JavaScript 等多种语言读取
• Schema 演进：支持 Schema 变更，便于数据结构升级

技术规格

属性	值
压缩算法	Snappy
文件格式	Apache Parquet
依赖库	pyarrow >= 14.0.0

数据分区存储结构

批量拉取的数据按照不同维度进行分区存储，便于数据管理和查询。

目录结构

data/
├── market/                    # 行情数据
│   ├── 000001/               # 股票代码
│   │   ├── 2024-01.parquet   # 按月分区
│   │   ├── 2024-02.parquet
│   │   └── ...
│   ├── 000002/
│   └── ...
├── north_flow/               # 北向资金数据
│   ├── 2024-01.parquet      # 按月分区
│   ├── 2024-02.parquet
│   └── ...
├── lhb/                      # 龙虎榜数据
│   ├── 2024-01-15.parquet   # 按日分区
│   ├── 2024-01-16.parquet
│   └── ...
├── sentiment/                # 舆情数据
│   ├── 000001/              # 股票代码
│   │   ├── 2024-01-15.parquet  # 按日分区
│   │   └── ...
│   └── ...
└── financial/                # 财报数据
    ├── 000001/              # 股票代码
    │   ├── 2024.parquet     # 按年分区
    │   ├── 2023.parquet
    │   └── ...
    └── ...

分区策略

数据类型	分区维度	文件命名规则	说明
market	股票代码 + 月份	{symbol}/{YYYY-MM}.parquet	按股票代码和月份分区
north_flow	月份	{YYYY-MM}.parquet	按月份分区
lhb	日期	{YYYY-MM-DD}.parquet	按日期分区
sentiment	股票代码 + 日期	{symbol}/{YYYY-MM-DD}.parquet	按股票代码和日期分区
financial	股票代码 + 年份	{symbol}/{YYYY}.parquet	按股票代码和年份分区

数据去重策略

在增量拉取时，系统会自动处理重复数据：

1. 读取现有数据：读取已存在的 Parquet 文件
2. 合并新数据：将新数据与现有数据合并
3. 去重处理：根据主键字段去重，保留最新数据
4. 覆盖写入：将合并后的数据写回文件

主键字段

数据类型	主键字段
market	date
north_flow	date
lhb	date, symbol
sentiment	date, title
financial	report_date

注意事项

1. 确保 Python 版本 >= 3.11
2. 确保已正确安装所有 Python 依赖（akshare、pandas、pyarrow）
3. 网络连接正常，以便 akshare 获取实时数据
4. 股票代码格式：上海市场以 6 开头，深圳市场以 0 或 3 开头
5. 批量拉取耗时较长，建议在非交易时间执行
6. 首次批量拉取会获取历史数据，耗时可能超过 1 小时
7. 增量拉取仅获取新数据，耗时较短