Stock News Radar

Problem

Watchlist signals are scattered across announcements, finance portals, and industry news. Manually checking sources, tracking individual tickers, and writing a pre-market brief can be slow and noisy, especially when self-published content is mixed in. Stock News Radar puts watchlist maintenance, monitoring, and briefing generation into one repeatable workflow for engineers or investors tracking A-share and ETF message flow.

Workflow and boundaries

The skill treats 自选股列表_标签.md as the single source of truth, supports adding or editing tickers by text, screenshot, or batch input, and keeps industry/concept labels attached to each entry. When running analysis, it reads the watchlist, pulls an Eastmoney announcement pool for breadth, then searches macro, industry, and ticker-specific news under domain restrictions; blocked_domains filters sources such as Baidu Baijiahao, Sohu, Xueqiu, and stock forums. Each item should show the source domain and separate factual announcements from inference. Output follows a V3 briefing template with an impact rating from major positive to major negative, and the quick table must cover every watchlist item. It focuses on message flow rather than technicals, valuation, or DCF, and results depend on API access, search quality, and label coverage.