What it does

Awesome-finance-skills is a collection of eight plug-in modules—“skills” in agent-framework parlance—that fetch real-time news, stock data, sentiment scores, and even generate research reports. You install individual skills via npx skills add or copy them into your agent’s skills directory. Each skill is a self-contained folder with a SKILL.md file that frameworks like OpenCode, Claude Code, or Antigravity pick up automatically.

The interesting bit

The project treats financial analysis as an agent capability layer rather than a standalone app. The logic-visualizer skill spits out Draw.io XML diagrams explaining market transmission chains—essentially giving your agent a whiteboard to show its work. The Kronos predictor claims news-aware adjustments, though the README doesn’t explain what Kronos is or how the adjustment works.

Key highlights

• alphaear-news: Aggregates 10+ sources including Cailian, WSJ, Weibo, and Polymarket
• alphaear-stock: A-share, HK, and US tickers with OHLCV and fundamentals
• alphaear-sentiment: FinBERT/LLM scoring on a -1.0 to +1.0 scale
• alphaear-reporter: Full pipeline from planning through chart generation
• alphaear-search: Web search plus local RAG via Jina, DuckDuckGo, or Baidu
• One-step install per-skill via npx skills; manual clone-and-copy fallback

Caveats

• The “Kronos model” is namedropped without citation or technical detail
• No visible tests, CI, or code structure in the README—it’s essentially documented glue
• Live demo links to external services (opncd.ai, deepear.vercel.app) with no guarantee of uptime

Verdict

Worth a look if you’re already running Claude Code, OpenCode, or similar and want pre-built finance primitives without wiring APIs yourself. Skip it if you need transparent models, audited predictions, or anything resembling a trading system you’d actually bet on.