返回 Skill 列表
extension
分类: 开发与工程无需 API Key

stock-data-sourcing

选择并结合市场数据和新闻提供商,用于A股、港股和美股的分析。当Codex需要在efinance、akshare、tushare、pytdx、baostock、yfinance、Tavily、Bocha、Brave或SerpAPI之间做出选择时使用;解释各提供商之间的权衡;设计回退链;或规划跨市场的观察列表和报告的数据摄入。

person作者: jakexiaohubgithub

Stock Data Sourcing

Overview

Use this skill to route cross-market data requests to the right providers before writing analysis or building ingestion workflows.

Prefer market coverage and freshness over provider purity, and always state the fallback chain when source quality is uncertain.

Workflow

  1. Classify the request by market and task:
    • market: A-share, Hong Kong, US, index, ETF
    • task: realtime quote, daily history, chips, fundamentals, market breadth, news, event intel
  2. Read references/data-sources.md when choosing or justifying providers.
  3. Pick a primary provider and at least one fallback.
  4. If marketbot already has a native tool that covers the task, use it and keep this skill as routing guidance.
  5. In the answer, state:
    • chosen source
    • fallback chain
    • freshness caveat
    • gaps that still need integration

Routing Rules

Quotes and history

  • A-share:
    • prefer tushare when a token is configured and stable structured data matters
    • otherwise prefer efinance for broad free coverage
    • fall back to akshare, then pytdx, then baostock
  • Hong Kong:
    • prefer akshare for free historical and realtime HK support
    • fall back to yfinance when symbol conversion is acceptable
  • US stocks and US indices:
    • prefer yfinance
    • do not route US history through akshare or efinance

Specialized enrichments

  • chips / cost distribution: prefer akshare
  • fundamentals / boards / company profile: prefer efinance
  • market indices / sector rankings:
    • A-share: efinance, akshare, or tushare
    • US: yfinance for indices; do not over-promise sector breadth unless a provider is actually wired

News and event intel

  • A-share and Chinese news:
    • prefer Bocha, then Tavily, then SerpAPI
  • US and global English news:
    • prefer Brave, then Tavily, then SerpAPI
  • Default freshness target: keep news within 3 days unless the user asks for a longer window.
  • For catalyst work, search across these dimensions:
    • latest news
    • market analysis / analyst view
    • risk check
    • earnings
    • industry context

Pairing

  • Use market-report for the final write-up.
  • Use catalyst-tracker when the user cares about drivers and calendars.
  • Use stock-info-explorer for Yahoo-based charts after the source plan is decided.
  • Use risk-checklist after the source-backed thesis is stable.

Output pattern

# Source Plan: <SYMBOL or WATCHLIST>

## Primary
- Provider:
- Why:

## Fallbacks
- 1:
- 2:

## Freshness / Limits
- Expected latency:
- Coverage gaps:
- News window:

## Integration Notes
- Existing marketbot tools to use:
- New connectors still needed:

Rules

  • Do not claim a provider is already integrated into marketbot unless the code path exists.
  • Separate current runtime capability from recommended future integration.
  • Prefer exact provider names over vague phrases like "some API".
  • If coverage is mixed, explicitly say which asset classes are strong and which are weak.