Investment analysis agent skills — options flow scanner + price alerts (open source Claude Code)

[tellmefrankie]

Hey deepagents community,

Deep agent research and implementation — sharing an open source project that demonstrates practical agentic investment analysis.

AI Investment Skills — Claude Code agent skills:

• Options flow scanner — sector ETF P/C anomaly detection with volume quality filtering
• Price monitor — scheduled monitoring, stop-loss/take-profit with Telegram alerts
• Daily briefing — multi-source synthesis

Agent architecture used:
The options scanner follows a ReAct-style loop:

1. Observe: Fetch current options chain data
2. Reason: Filter junk volume, compute adjusted P/C
3. Act: If anomaly detected, send Telegram alert
4. Reflect: Log signal for historical pattern building

Real signal:
XLI P/C hit 5.32 (normal: 0.5–1.2). After filtering 98% lottery ticket volume on CEG: "extremely bearish." The filtering step is the key reasoning step.

Open source:
→ https://github.com/tellmefrankie/ai-investment-skills

Anyone here using deepagents for financial domain applications?