daily-stock-analysis
LLM-powered A/H/US stock intelligent analysis system with multi-source data, real-time news, AI decision dashboards, and multi-channel push notifications via GitHub Actions.
Security Assessment
Detected risks:
About daily-stock-analysis
Daily Stock Analysis is an LLM-powered analysis system for A-share, Hong Kong, and US stocks that fetches quotes, news, and fundamentals, generates an AI decision dashboard, and pushes the results to messaging channels on a schedule. Each stock's dashboard distills to a one-line conclusion plus precise buy, sell, and stop-loss prices and a checklist. It covers multiple markets and indices (including SPX, DJI, IXIC), pulling quotes from AkShare, Tushare, and YFinance and news from Tavily, SerpAPI, or Brave.
Model access is unified through LiteLLM across backends such as Gemini, OpenAI, Claude, DeepSeek, and Qwen, and reports can be delivered to WeChat Work, Feishu, Telegram, Discord, DingTalk, Email, and PushPlus. The headline deployment path is GitHub Actions, which runs the analysis on a cron schedule at zero server cost: fork the repository, configure secrets (at least one LLM key, a comma-separated STOCKS list, and at least one notification channel), then trigger the workflow manually or wait for the schedule. Local and Docker installation are also supported via a .env file, with Docker and Docker Compose variants.
Configuration is extensive. Beyond keys and the stock list, options include report type, an inter-stock analysis delay to avoid rate limiting, worker concurrency, per-stock immediate notifications, and news age filtering. Advanced setups allow multiple LLM channels, stock grouping to route different tickers to different email recipients, and a market review mode for A-share or US regime strategies. A web dashboard (python web_app.py) adds portfolio and P&L tracking, history, backtesting, and an Agent Q&A with 11 built-in strategies such as MA crossover and Elliott Wave. Documented code examples show programmatic single-stock analysis returning conclusion and target prices, custom Telegram and Feishu notifications, an agent chat API, backtesting that reports direction accuracy and take-profit/stop-loss hit rates, and importing a watchlist from a screenshot via a vision model. The GitHub Actions schedule is defined in .github/workflows/stock_analysis.yml, and secrets can be set with the gh CLI.
FAQ
Which markets and data sources does it cover?
A-shares (CN), Hong Kong, and US stocks plus indices like SPX, DJI, and IXIC, using AkShare, Tushare, and YFinance for quotes and Tavily, SerpAPI, or Brave for news.
What do I need to configure to run it on GitHub Actions?
Fork the repo and set secrets for at least one LLM key, a comma-separated STOCKS list, and at least one notification channel; then trigger the workflow manually or let the cron schedule run it.
Which LLM backends and notification channels are supported?
LLM backends include Gemini, OpenAI, Claude, DeepSeek, and Qwen via LiteLLM; push channels include WeChat Work, Feishu, Telegram, Discord, DingTalk, Email, and PushPlus.
Can I run it without GitHub Actions?
Yes. It supports local installation via a .env file and python main.py, as well as Docker and Docker Compose.
Does it offer backtesting?
Yes. The web dashboard and a backtest API can evaluate recent AI predictions, reporting metrics such as direction accuracy and take-profit/stop-loss hit rates.
Install daily-stock-analysis
Quick Setup:
- Copy the skill folder to
.claude/skills/ - Claude will automatically detect and use the skill
Repository
aradotso/trending-skills