Vibe-Research

# Tool List [](LICENSE) [](https://www.python.org/) [](https://react.dev/) [](https://fastapi.tiangolo.com/) [](https://github.com/simonlin1212/Vibe-Research/stargazers) [](https://viberesearch.wiki) **[Official Website](https://viberesearch.wiki) · [Product Preview](#product-preview) · [Features](#features) · [Data Sources](#data-sources) · [Quick Start](#quick-start) · [AI Integration](#ai-integration) · [Compliance](#compliance) · [Related Ecosystem](#related-ecosystem) · [Contact Author](#contact-author)** > **Vibe-Research: Your Personal Trading Research Agent** · A personal trading research agent for A-share, US stock, and Hong Kong stock. > Daily review, news radar, stock data, sector center, my holdings, research records. Equip with data and features, driven by **your own AI**. Vibe-Research is an open-source personal AI trading research panel, **mainly for A-share, also for US / HK stock** (A-share often needs to watch overnight foreign markets, with more complete data). It does not make decisions for you - it provides market data, research reports, valuations, financial data, announcements, capital flows, and news, all in a clean panel, and leaves an interface for **your own AI**. > *Vibe-Research: Your Personal Trading Research Agent. An open dashboard for China A-share (plus US / HK): it wires up all the data and plugs into **your own AI / agent** — it never recommends a stock. You bring the model, it brings the data.* ## 🖥️ Product Preview **Daily Review** — A-share / short-term sentiment (continuous board stocks · top 20 trading volume) / sector fund flow, all in one screen, one-click review by your AI  <table> <tr> <td width="50%"> **Stock Data** — financial report overview + valuation percentile + capital flow, all in one screen  </td> <td width="50%"> **News Radar** — 12 tracks, 108 public sources, one-click extract today's key points  </td> </tr> </table> ## ✨ Features Specific modules for each page: | Page | Modules / Capabilities | |---|---| | 📊&nbsp;**Daily Review** | A-share index · **Global Market** (overnight US stock market / Hang Seng Index) · watched stocks (real-time quotes) · **Short-term Sentiment** (continuous board stocks / highest continuous board / board ladder / sealed board rate / exploded board rate / promotion rate) · **Top 20 Trading Volume** · Market Sentiment (market breadth / theme speculation / limit up / limit down) · Sector Fund Flow · AI Daily Review | | 📡&nbsp;**News Radar** | 12 tracks, 108 public RSS sources · AI one-click extract "Today's Key Points" · A-share announcements / public news (linked to your watchlist) | | 🔍&nbsp;**Stock Data** | **A-share**: quotes · valuation matrix (forward PE / PEG) · **financial report overview** · valuation historical percentile · financial key indicators · research reports · announcements · news · **capital flow** (financing and securities lending / shareholder number / main capital flow / dividend / block trading) · Dragon and Tiger List · restricted stock lifting · sector affiliation · popular concepts · interactive Q&A. **US / HK stock**: quotes · total market value · key financial indicators (revenue / net profit / EPS / ROE / gross margin / debt ratio) | | 🧩&nbsp;**Sector Center** | sector + industrial chain link skeleton | | 💼&nbsp;**My Holdings** | input and real-time profit and loss · cleared record (only stored locally, not uploaded) | | 📝&nbsp;**Research Records** | review / today's key points / AI Q&A results, locally stored,随时回看 | | 🔌&nbsp;**AI Integration** | subscription integration (local CLI, no key) · API multi-model (automatic baseURL) · MCP (connected to Claude Code, etc.) | > **Investment Research Framework**: let AI analyze stocks, automatically organized by valuation / capital flow / financial quality / industry prosperity / event catalysis and risk five-dimensional conclusions — framework only specifies "how to read data", not buy or sell, direction still determined by **your own AI**. ## 📡 Data Sources Vibe-Research directly integrates three sets of public data sources **into the repository** — `git clone` and **ready to use, no additional download or connection**. ### A-share Full-Stack Data · AStockData - **In the repository's [`a-stock-data/`](a-stock-data/) folder** (v3.3.0). Ten-layer data architecture, 40 endpoints, `a-stock-data/SKILL.md` **embedded with all calling code**, self-contained, zero third-party data packaging dependencies, Dongcai interface with built-in current limit anti-sealing. - **Coverage**: quotes / K-line / research reports / consensus expectations / valuations / historical percentiles / financial statements / announcements / Dragon and Tiger List / financing and securities lending / block trading / shareholder number / dividend / capital flow / lifting restrictions / concept sectors / playing board sentiment / ETF options / interactive Q&A / full market industry ranking ... - **For agent use**: when running this repository with Claude Code, etc., to call A-share data, see [`a-stock-data/SKILL.md`](a-stock-data/SKILL.md) — each interface has copy-paste ready-to-use code. Vibe-Research backend data layer (`backend/astock.py`) is also transplanted from it. - **Runtime dependencies**: `pip install mootdx requests pandas stockstats` (self-contained, v3.0 and above removed akshare dependencies). - **Update / upstream**: <https://github.com/simonlin1212/a-stock-data> — want to keep up with the latest endpoints, expand data sources, go here; **but even if you don't update, the repository's built-in snapshot is fixed and usable, can be used forever**. ### US / HK Stock Data · global-stock-data - **In the repository's [`global-stock-data/`](global-stock-data/) folder** (v1.0.1). Eight-layer data architecture, 18 endpoints, zero authentication, covering US and HK stock quotes / K-line / technical indicators / financial statements / capital flow / options / SEC. - Backend `backend/gstock.py` transplanted **Dongcai domain compliant subset**: global index (daily review "Global Market" column) + US and HK stock quotes & key financial indicators (stock page input `AAPL` / `00700` to use). Dongcai call reuse `astock.em_get` (direct connection priority, avoid scientific internet proxy domestic station). - **Upstream**: <https://github.com/simonlin1212/global-stock-data> — want K-line / technical indicators / options / SEC and other full-endpoint, go here. ### Global News · investment-news - 12 tracks, 108 public RSS sources, integrated into `backend/newsradar.py` + `backend/news_sources.json`: pure standard library, zero key, filtered by compliance vocabulary (excluding gambling / prediction market / encryption, etc.). - **Upstream**: <https://github.com/simonlin1212/investment-news> ## 🏗 Architecture One data layer + two AI exits: ``` Vibe-Research/ ├── a-stock-data/ A-share full-stack data toolbox (data source, v3.3, ready to use) ├── global-stock-data/ US / HK stock data toolbox (data source, v1.0.1, ready to use) ├── backend/ FastAPI :8900 │ ├── astock.py A-share data (transplanted from a-stock-data) │ ├── gstock.py US / HK stock data (transplanted from global-stock-data) │ ├── newsradar.py News Radar (transplanted from investment-news) │ ├── market.py market sentiment + sector capital flow + global index │ ├── portfolio.py holdings + cleared (local cache) │ ├── chat.py system AI (OpenAI compatible function-calling) │ └── mcp_server.py MCP server (for Claude Code, etc.) └── frontend/ Vite + React 19 + TS + Tailwind (glass orange theme):5899 ``` **Hierarchical dependencies**: quotes (Tencent) + research reports / announcements (Dongcai) **installed and available in seconds**; akshare / mootdx lazy import, missing endpoint returns 501 + installation prompt, does not drag down service. ## 🚀 Quick Start ```bash # Backend (:8900) cd backend && python3 -m venv .venv && .venv/bin/pip install -r requirements.txt .venv/bin/python -m uvicorn app:app --host 127.0.0.1 --port 8900 # Frontend (:5899) cd frontend && npm install && npm run dev # Open browser and navigate to http://localhost:5899 ``` ## 🔌 AI Integration Configure once on the "AI Integration" page, and all "AI Q&A / Daily Review / Today's Key Points" will use **your own model**. **Analysis is provided by your model, this product does not calibrate or bias**. Three ways: ### 1. Subscription Integration (call local CLI, no API key) Use **your own subscription quota**, no API fee. Supported: **Claude Code · Codex · Qwen Code · Gemini CLI · DeepSeek CLI**. - **Prerequisites**: ① Backend runs on your local machine (cloud cannot read your local CLI); ② corresponding CLI installed and logged in, command in `PATH`. For example: - Claude Code: `npm i -g @anthropic-ai/claude-code` → `claude` (login with Claude subscription) - Codex: install OpenAI Codex CLI → `codex login` (login with ChatGPT subscription) - Qwen / Gemini / DeepSeek: install respective CLI and log in - Select one on the "AI Integration page → Subscription Integration", **no key required**. - Principle: backend `cli_runtime.py` detects local commands and `spawn` it to answer once (data already in prompt). ⚠️ CLI does not perform multi-turn tool calls, suitable for **data-ready** scenarios like "Daily Review / Today's Key Points / Stock Page AI Q&A". ### 2. API Integration (fill in your own key) Select a model on the "AI Integration page → API Integration", **baseURL automatically filled**, just paste the key. Built-in **DeepSeek / 豆包 / MiniMax / OpenAI / OpenRouter / Groq / Together / MiMo / any OpenAI compatible endpoint**. This option supports function-calling — AI will call data tools (quotes / valuation / research reports / news) and then answer. key is stored locally in your browser, sent with requests to your backend, **not uploaded, not into repository**. ### 3. MCP (for Claude Code / advanced agent) Make backend an MCP server, agent uses **your own subscription quota** to call Vibe-Research data tools, multi-step analysis. Command see [`backend/README.md`](backend/README.md). For more A-share data endpoints, use root directory [`a-stock-data/`](a-stock-data/SKILL.md) toolbox. ## 🧪 Testing ```bash cd backend && .venv/bin/pip install -r requirements-dev.txt .venv/bin/pytest -m "not live" # offline unit test + API verification (fast, stable, no internet required) .venv/bin/pytest -m live # online verification of data source shape (upgrade / release before running) ``` ## ⚖️ Compliance - Only objective data sorting and public list presentation: **no stock recommendation, no price prediction, no trading timing, no return commitment, no subjective scoring**; neutral and unbiased. - Continuous board stocks / trading volume rankings are **objective public list data** (Dongcai / Tongdun same data), product only presents facts, **no recommendation or prediction**. - All analysis direction provided by **your own AI**, unrelated to this product. UI has no buy/sell button; valuation historical percentile only marks position, **no trading line**. - **Holdings / watched stocks / API key stored locally in browser, not uploaded, not into repository**. ## 🏛 Related Ecosystem Data / tools used by Vibe-Research come from the same self-developed open-source system (all in [`simonlin1212`](https://github.com/simonlin1212)): | Repository | Positioning | |---|---| | [**a-stock-data**](https://github.com/simonlin1212/a-stock-data) | A-share full-stack data toolbox (10 layers · 40 endpoints) — A-share data engine for this project | | [**global-stock-data**](https://github.com/simonlin1212/global-stock-data) | US / HK stock full-stack data toolbox (7 layers · 17 endpoints) | | [**investment-news**](https://github.com/simonlin1212/investment-news) | global industry news dashboard (12 tracks corresponding to A-share sectors) — news source for this project | | [**Agent-Staff**](https://github.com/simonlin1212/Agent-Staff) | make company agent: each department an AI agent + CEO chief of staff, permanent Feishu | ## 📬 Contact Author Author **Simon**, independent developer. - 🌐 Homepage: <https://www.simonlin.net> - 💬 Welcome to discuss **enterprise AI solutions**; project-related issues can also be raised in [Issue](https://github.com/simonlin1212/Vibe-Research/issues). ## 🙏 Acknowledgements - A-share data engine: [a-stock-data](https://github.com/simonlin1212/a-stock-data) (author: Simonlin1212) - US / HK stock data engine: [global-stock-data](https://github.com/simonlin1212/global-stock-data) (author: Simonlin1212) - News: [investment-news](https://github.com/simonlin1212/investment-news) (author: Simonlin1212) - Interface design language reference and thanks: [HKUDS/Vibe-Trading](https://github.com/HKUDS/Vibe-Trading) (author: HKUDS · only UI reference, underlying fully implemented) ## ⚠️ Disclaimer This project is for learning and research only, **does not constitute any investment advice**. The dashboard only sorts objective data and presents public lists — **no stock recommendation, no price prediction, no trading timing, no return commitment**; all analysis direction provided by **your own AI**, unrelated to this product. Stock market has risks, please make independent decisions, verify yourself, and bear risks. ## 📄 License MIT