A Claude Code skill that turns Claude into an autonomous financial research analyst. Ask it to analyze, value, compare, or screen public companies and it pulls real market data — fundamentals, prices, valuation ratios, news, analyst forecasts, insider trades, and institutional holdings — then reasons over it.
It ships with three interchangeable data backends so you're never blocked:
| Backend | Script | Coverage | API key | Best for |
|---|---|---|---|---|
| yfinance | scripts/yf.py |
Whole market (incl. small/mid caps) | none | Default — works for any ticker, free |
| Financial Datasets | scripts/fd.py |
US; free tier = a few large caps | yes | LLM-clean format |
| Finnhub | scripts/finnhub.py |
Whole market | free key | Analyst estimates, recommendation trends |
The skill cross-validates figures across backends when it matters (price, multiples, margins).
This is a personal Claude Code skill. Clone it into your skills directory:
git clone https://github.com/<you>/alphascout.git \
~/.claude/skills/alphascout
cd ~/.claude/skills/alphascout
./install.sh # creates .venv and installs yfinanceThe yfinance backend works immediately with no API key. For the other two:
cp .env.example .env # then add your keys (or export them as env vars)- Financial Datasets key: https://financialdatasets.ai
- Finnhub free key: https://finnhub.io/register
Just talk to Claude Code naturally — the skill triggers automatically:
- "analyze NVDA's last 5 years of revenue and margins"
- "compare AAPL, MSFT and GOOGL on gross margin and FCF"
- "is NU overvalued? what's the upside to analyst targets?"
- "research the memory sector — MU, SK Hynix, Sandisk"
Or run a backend directly:
.venv/bin/python -W ignore scripts/yf.py income AAL --limit 5
.venv/bin/python -W ignore scripts/yf.py forecast NU
python3 scripts/finnhub.py recs MU
python3 scripts/fd.py income AAPL --limit 4You: is NU overvalued? what's the upside?
Claude (via alphascout) pulls fundamentals + analyst data from yfinance and cross-checks the multiples and recommendation trend against Finnhub, then answers:
NU Holdings (NU) — $13.05, mktcap $63.4B. Cross-validated yfinance × Finnhub:
Metric yfinance Finnhub P/E (TTM) 20.1 19.9 ROE 30.1% 28.9% Rev growth +43.7% +45.4% Revenue grew 3.5× in three years (loss → 27% net margin); ROE ~30% is top-tier for a bank. Forward P/E ~11 with EPS forecast to keep compounding ~30%/yr → not expensive for the growth. Analyst consensus Strong Buy, mean target $19.39 (+49%), range $15.3 (+17%) to $22 (+69%). Trades near its 52-week low despite strong fundamentals.
Upside: base case ~+50%; bull ~+70–85%; bear +17%. Main risk is Brazil/Mexico macro (rates, FX, credit), not the franchise. Educational only — not financial advice.
Under the hood that was roughly:
.venv/bin/python -W ignore scripts/yf.py metrics NU
.venv/bin/python -W ignore scripts/yf.py forecast NU
python3 scripts/finnhub.py metrics NU # cross-check multiples / margins
python3 scripts/finnhub.py recs NU # recommendation trendThe skill picks the backend, batches the calls, reconciles any disagreements between sources, and leads with the conclusion — you just ask the question.
For educational and informational purposes only. Not financial advice. Data may be incomplete, delayed, or wrong — always verify before making any decision. yfinance scrapes Yahoo Finance and is not an official API.
- Methodology & design inspired by Dexter by @virattt — an autonomous financial research agent (MIT License). This skill reimplements the data-tool + research-workflow idea natively for Claude Code; it does not include Dexter's source code.
- Data sources: Financial Datasets, yfinance (Yahoo Finance), and Finnhub.