a-stock-data

// Overview

a-stock-data is a self-contained Skill file — structured Markdown with embedded Python — that gathers raw China A-share and related market data from 34 sources into a toolkit an AI coding assistant can use directly. The project describes its scope as 15 layers, 87 capability endpoints (82 primary plus 5 backups) and 34 data sources, and every source except iwencai's natural-language research search works without a user account or API key.

The layers cover quotes and K-lines, broker research reports and PDFs, market signals such as northbound flow and the dragon-tiger board, margin and capital-flow data with locally computed chip distribution, news feeds, fundamentals and financial statements, exchange filings, limit-up pools, ETF options with Greeks and implied volatility, investor Q&A sentiment, macro and interest-rate series, index constituents and the trading calendar, futures and commodities, event-driven data (earnings previews, buybacks, share pledges, IPO calendar) and convertible bonds. The point is that the assistant no longer has to know Tencent's K-line paging parameters, the binary layout of TDX end-of-day packages or Eastmoney's PDF Referer header — the skill wraps them.

It is built around source reliability. Sources are ranked by IP-ban risk — Tencent and official exchange data first, Eastmoney only for data nobody else has, routed through a throttled `em_get()` helper — and a backup-source reference lists independent fallbacks for when a primary source blocks you. The skill only fetches data: it has no backtesting engine and gives no investment advice. Keeping it as one file is a deliberate product decision, so the author suggests token-sensitive users load it on demand and use its endpoint routing table to read only the layer they need.

// What it does

• 15 data layers and 87 endpoints spanning quotes, K-lines, tick trades, research, fund flow, news, financials, filings, options, macro, futures, events and convertible bonds
• Keyless by design: every integrated source except iwencai's natural-language search needs no registration or API key
• Source priority by IP-ban risk, with Eastmoney calls serialised through a rate-limited `em_get()` helper
• Backup-source quick reference with independent fallbacks for the dragon-tiger board, fund flow, filings, margin data and Beijing Stock Exchange quotes
• Works in Claude Code as an auto-triggered skill, or in Codex, OpenClaw and other assistants by pasting it into context
• Offline unit tests extracted from SKILL.md itself, plus opt-in live network tests per release

// Getting started

Installation is three steps: create a skill directory, download SKILL.md into it, and install the Python dependencies. Only iwencai's research search needs an API key.

mkdir -p ~/.claude/skills/a-stock-data

curl -o ~/.claude/skills/a-stock-data/SKILL.md \
  https://raw.githubusercontent.com/simonlin1212/a-stock-data/main/SKILL.md

pip install mootdx requests pandas stockstats numpy baostock xlrd openpyxl

Commands and code are distilled from the project's own documentation — always check the official repo for the latest.

// When to use it

• Let a coding assistant value an A-share stock from live price, consensus EPS and forward PE without writing scrapers
• Pull limit-up pools, dragon-tiger seats and sector fund flow into an agent's daily market recap
• Feed research reports, filings and investor Q&A into an agent that investigates a new ticker
• Collect adjusted K-lines, valuation history and industry history as input for your own backtesting framework

// How a-stock-data compares

a-stock-data alongside other open-source finance & market agents tools AI/TLDR tracks, ranked by GitHub stars.