@AYi_AInotes: Wow, these two GitHub projects must be recommended together. People doing AI investment research can save months of effort. Someone turned the full free data of A-shares + US and Hong Kong stocks into an AI-native Skill. No need to integrate APIs, no need to handle anti-scraping, almost zero API keys. In Claude, Cursor, Codex…

Holy crap, these two GitHub projects have to be recommended together.

AI investment research folks can save months of effort now—someone has turned the full free data for A-shares + US/HK stocks into an AI-native Skill.

No need to integrate APIs, no dealing with anti-scraping hassles, almost zero API keys required. One line in Claude Cursor Codex pulls up market quotes, research reports, fund flows, dragon/tiger lists.

In other words, one covers full-dimensional A-share data, the other unlocks US/HK stocks + options chains. After installing Claude Code, one line pulls market quotes, research reports, fund flows—zero API keys.

This thread covers the two open-source projects, plus a killer combo with the UZI analysis engine.

Let’s start with the main A-share project: a-stock-data.

7-layer architecture, 28 endpoints, directly connecting to 13 data sources. After v3.0, the third-party middleware was completely removed—everything goes through native HTTP/TCP interfaces, maximizing anti-ban stability. It covers seven core dimensions: Real-time K-line valuation, research report PDF consensus expectations, Dragon/tiger list, northbound capital flows, margin trading, block trading, Earnings reports, announcements, news, lock-up expirations, and even natural language thematic research report search. Built-in 4 automated research workflows: single-stock valuation, batch comparison, new target research triggered with one sentence. Installation is three steps. Compatible with all AI coding platforms.

The sister project global-stock-data covers US/HK stocks and is more lightweight.

It only relies on the requests library, with 5 data sources and 18 endpoints—all zero authentication. Its most competitive feature is deep US stock capabilities: Full options chain with complete Greeks, SEC EDGAR fetching 503 structured GAAP indicators, Pure Python implementation of the full technical indicator suite, no extra dependencies. US stock K-line data goes back to 1984, full-dimensional HK stock coverage. It can be installed alongside the A-share project without conflicts, giving you one-click access to the A-share, HK, and US markets.

The real killer move is stacking UZI-Skill on top to form a complete closed loop.

The two data projects handle all the dirty work of raw data acquisition, while UZI-Skill handles the intelligent analysis layer: 66 investment judges + 22 data dimensions + 17 institutional analysis models, directly outputting Bloomberg-style structured HTML reports. From pulling raw data, to calculating indicators, to multi-expert debate, to generating a full report— the entire process is triggered by a single sentence. Even a regular person can quickly build a complete AI quantitative research system.

Both projects are completely open source and free, with no paywalls.

For anyone doing quantitative trading, building AI agents, or creating financial content, this setup saves at least months of work on data source integration and maintenance. Project links are in the comments. Suggest starring them first, and you can tinker over the weekend. What strategies do you want to try with these tools?

Haha, I’ll walk you through it step by step—it all depends on your understanding and execution.