If you've ever wanted a desktop app that reads stock news for you and tells you — with AI — whether it's bullish or bearish, StockAI is exactly that. I'm excited to share this open-source project with the DEV community!
What is StockAI?
StockAI is a cross-platform desktop application built with Tauri 2, React 19, Bun, and Playwright. It scrapes real-time stock news (US stocks + Chinese A-shares), runs the articles through your choice of LLM, and presents a clean sentiment analysis dashboard with interactive K-line charts.
Think of it as your personal AI financial analyst running 100% locally — no subscriptions, no data leaks, just insights.
🌟 Key Features
1. Pluggable AI Providers — Switch with a Dropdown
StockAI supports five AI providers out of the box:
- OpenAI (GPT-4o and friends)
- Anthropic (Claude 3.5 Sonnet)
- DeepSeek (DeepSeek V4 Pro)
- GLM (GLM-5.1)
- Ollama (any local model — Llama 3, Mistral, Qwen, etc.)
Each provider stores its own API key, base URL, and model selection. You can switch between them via a single dropdown. And importantly: analysis is explicitly triggered, meaning switching stocks never silently burns your tokens.
2. Real-Time News Scraping
News is fetched via Google News RSS (no heavy Chromium boot required for most stocks). If RSS doesn't cut it, Playwright takes over as a fallback for deeper scraping.
It supports:
- US stocks (NYSE, NASDAQ)
- Chinese A-shares: Shanghai (SH), Shenzhen (SZ), and Beijing (BJ) Stock Exchanges
3. Interactive K-Line Charts (Powered by Lightweight Charts)
The central panel features a full-featured candlestick chart with:
- MA / BOLL overlays
- 9 time periods: 1D / 5D / 1M / 3M / 6M / YTD / 1Y / 5Y / All
- Sub-chart indicators: MACD, RSI, KDJ, OBV, VWAP
- Logarithmic scale, dividend-adjusted prices
- Real-time price merged into the last candle during trading hours
4. Three-Column Dashboard Layout
The dark-themed UI is split into three columns:
- Left: Watchlist sidebar with quick-add input (persists across sessions)
- Center: Stock detail panel — real-time price card, K-line chart, news feed
- Right: AI analysis panel — sentiment ratio bar, company profile, color-coded bullish/bearish factor cards
5. Local-First Privacy
All your API keys, watchlists, and settings live locally on your machine. Nothing is sent to any StockAI server (there isn't one).
6. Deep Mode
Enable "Deep Mode" to have the scraper extract the full article content (not just headlines), giving the LLM far more context for richer sentiment analysis.
🏗️ Architecture
StockAI follows Clean Architecture with a strict unidirectional dependency flow:
UI (React + TypeScript + Vite)
↓
Core Orchestration (Tauri / Rust)
↓
Analysis Engine (Bun Sidecar + Playwright + AI SDKs)
The Bun sidecar is compiled into a standalone binary, managed by Tauri's sidecar mechanism. This keeps the frontend fast and the heavy scraping/AI work isolated.
🚀 Getting Started
Download (Recommended)
Pre-built binaries for macOS, Windows, and Linux are on the Releases page.
macOS users: If you see "damaged app" — this is just Gatekeeper. Run:
xattr -cr /Applications/StockAI.app
Windows users: Click "More info → Run anyway" on the SmartScreen dialog.
Build from Source
# Prerequisites: Bun + Rust
# 1. Install dependencies
bun install
# 2. Compile the sidecar (example for macOS ARM64)
bun build sidecar/index.ts --compile --outfile sidecar/stockai-backend-aarch64-apple-darwin
# 3. Start dev mode
bun tauri dev
🛠️ Tech Stack
| Layer | Technology |
|---|---|
| Desktop Shell | Tauri 2.0 (Rust) |
| Frontend | React 19, TypeScript, TailwindCSS 4 |
| Charts | Lightweight Charts, Lucide Icons |
| Scraper | Bun, Playwright, NodeHtmlMarkdown |
| AI SDKs | OpenAI, Anthropic, Ollama |
💡 Why I Built This
As someone interested in both AI and markets, I was frustrated by how scattered the workflow was: open a news site, read a few articles, try to gauge sentiment, check the chart on another tab. StockAI puts it all in one clean desktop app.
I also wanted something that works without vendor lock-in — if OpenAI's pricing goes up, I can switch to local Ollama models with zero friction.
🤝 Contributing
The project is fully open source under the MIT License. Contributions, issues, and stars are very welcome!
👉 GitHub: https://github.com/hyhmrright/StockAI
Whether you want to add a new AI provider, improve scraping coverage, add more chart indicators, or just fix a typo — PRs are welcome. The codebase is clean and well-structured, making it easy to jump in.
What's Next?
Some ideas on the roadmap:
- Price alert notifications (already in progress!)
- Stock screener with batch quant scanning
- More exchange support (Europe, Japan)
- Portfolio tracking integration
If you find StockAI useful, drop a ⭐ on GitHub — it really helps the project grow. And if you build something with it or have feedback, let me know in the comments below!
Happy trading 📈
💬 Quick question for you: Which AI provider would you use with StockAI — OpenAI, Claude, DeepSeek, or a local Ollama model? And what stock markets do you care about most (US, China, or somewhere else)? Drop it in the comments — your answer might shape the next feature!
