silentfleetkk/ai-market-pulse

AI Market Pulse turns a personal stock or crypto watchlist into a daily research report. It does not connect to a broker, place trades, or promise any returns, it is a research and monitoring tool that pulls market data, runs technical analysis, and produces reports you read yourself before making any decision. The core loop has four steps. You enter your holdings, either by typing symbols, adding portfolio positions with quantity and cost, or uploading a screenshot of a brokerage account that an AI model transcribes into editable rows. The tool then calculates technical factors like moving averages, RSI, MACD, and drawdown, plus a rules based score from 0 to 100 with plain language reasons, and compares each holding to a benchmark such as the S&P 500 or CSI 300. You review this through a signal overview, a risk board, and a report grounded question and answer feature. Finally, the tool can detect meaningful changes, send alerts through Telegram, Slack, Discord, or email, and publish a bilingual static website through GitHub Pages, so the whole thing can run with no server of your own. It supports stocks, ETFs, crypto, Yahoo Finance compatible tickers, and mainland Chinese A-share codes, and it can pull data from providers like yfinance, AkShare, Baostock, or Tushare with automatic fallback if one is unavailable. You can try it in about a minute with a built in offline demo that uses sample data and needs no API keys at all. For real use, setup involves cloning the repo, creating a Python virtual environment, and installing with pip, then running either a local web console or a command line tool to generate reports for your own symbols. Optional AI features, such as summarizing a portfolio in plain English or reading a brokerage screenshot, require your own OpenAI compatible API key, and the README notes you should redact account identifiers before uploading a screenshot since it is sent to that AI provider. The project is released under the MIT License and is written in Python.