---

name: google-finance description: "Track stock prices and company news from Google Finance on a schedule. Use when user wants to monitor stocks, get buy/sell recommendations, check price changes, follow company news, set up stock alerts, or track portfolio symbols. Triggers: 'track stock', 'watch AAPL', 'stock alert', 'buy or sell', 'stock news', 'Google Finance'. Default watchlist: NVDA, AAPL, META, GOOGL." homepage: https://finance.google.com user-invocable: true metadata: {"clawdbot": {"emoji": "📈", "requires": {"bins": ["python3", "curl"]}, "os": ["darwin", "linux", "win32"], "files": ["scripts/parse-stock.py"]}}

Stock Tracker

Monitor stocks on Google Finance, surface company news, and generate buy/sell signals on a schedule.

Quick Start

Add a stock to the watchlist:

/stock-tracker add AAPL
/stock-tracker add TSLA NVDA MSFT        ← multiple at once
/stock-tracker add 0700.HK               ← HK-listed stocks

Run an immediate check on all watched stocks:

/stock-tracker check

Check a single symbol:

/stock-tracker check AAPL

Show watchlist:

/stock-tracker list

Remove a stock:

/stock-tracker remove TSLA

Set up a recurring cron schedule:

/stock-tracker schedule                  ← interactive setup

---

Step-by-Step Workflow

1. Fetch Price Data (Google Finance)

Navigate browser to https://www.google.com/finance/quote/{SYMBOL}:{EXCHANGE}.

Extract the following fields from the page DOM:

If the DOM selector fails, fall back to scraping the visible text between price markers. See {baseDir}/references/data-sources.md for exchange suffix mapping and fallback sources.

2. Fetch Company News

On the same Google Finance page, scroll to the "News" section and extract the top 5–8 headlines with:

• Title
• Source
• Published time (relative, e.g. "3 hours ago")
• URL

Also search: https://www.google.com/search?q={COMPANY_NAME}+stock+news&tbm=nws&tbs=qdr:d for the past 24 hours of news.

3. Run Analysis

Apply the scoring framework in {baseDir}/references/analysis-framework.md to produce:

Symbol: AAPL
Price: $182.30  (+1.4% today)
Signal: BUY  [score: +6/10]
Confidence: MEDIUM

Key factors:
  ✅ Price above 50-day SMA (estimated)
  ✅ Volume 1.3× above average
  ✅ 2 positive news items in past 24h
  ⚠️  P/E 28.5 — elevated but within sector norm
  ❌ Within 3% of 52-week high (limited upside)

Recent headlines:
  • "Apple reportedly in talks with..." — Reuters (2h ago)
  • "iPhone sales beat estimates..." — Bloomberg (5h ago)

Recommendation: Consider buying on dips. Set stop-loss at 5% below current price.

4. Persist State

Store watchlist and last-seen prices in ~/.openclaw/workspace/stock-tracker-state.json.

Default watchlist (pre-loaded on first run if state file does not exist):

• NVDA:NASDAQ — NVIDIA
• AAPL:NASDAQ — Apple
• META:NASDAQ — Meta Platforms
• GOOGL:NASDAQ — Alphabet (Google)

Format:

{
  "watchlist": ["NVDA:NASDAQ", "AAPL:NASDAQ", "META:NASDAQ", "GOOGL:NASDAQ"],
  "lastChecked": "2026-03-03T09:00:00Z",
  "snapshots": {
    "NVDA:NASDAQ": { "price": 875.40, "change_pct": 2.1, "ts": "2026-03-03T09:00:00Z" },
    "AAPL:NASDAQ": { "price": 182.30, "change_pct": 1.4, "ts": "2026-03-03T09:00:00Z" },
    "META:NASDAQ": { "price": 512.60, "change_pct": -0.8, "ts": "2026-03-03T09:00:00Z" },
    "GOOGL:NASDAQ": { "price": 175.20, "change_pct": 0.5, "ts": "2026-03-03T09:00:00Z" }
  }
}

Load state at the start of every run. Compare new price against snapshots to compute Δ since last check.

5. Alert on Significant Moves

Emit a highlighted alert if any of the following thresholds are crossed:

• Price change > ±3% since last check
• Volume > 2× 30-day average
• Any headline contains keywords: earnings, merger, acquisition, SEC, lawsuit, recall, CEO, bankruptcy

---

Setting Up a Cron Schedule

Run the following to add a recurring job that checks stocks every weekday at market open (09:30 ET) and close (16:00 ET):

# Market open — 09:30 ET (UTC-4 during EDT)
openclaw cron add \
  --name "Stock Open Check" \
  --cron "30 13 * * 1-5" \
  --tz "America/New_York" \
  --session isolated \
  --message "Run /stock-tracker check and output a full report with buy/sell signals for all watched stocks." \
  --announce \
  --channel slack \
  --to "channel:REPLACE_WITH_CHANNEL_ID"

# Market close — 16:00 ET
openclaw cron add \
  --name "Stock Close Check" \
  --cron "0 20 * * 1-5" \
  --tz "America/New_York" \
  --session isolated \
  --message "Run /stock-tracker check and output end-of-day summary with buy/sell signals for all watched stocks." \
  --announce

For a simpler every-4-hours check (including after-hours):

openclaw cron add \
  --name "Stock Tracker" \
  --every 14400000 \
  --session isolated \
  --message "Run /stock-tracker check for all watched symbols. Report price changes > 1%, news, and signals."

See {baseDir}/references/data-sources.md for timezone and exchange hours reference.

---

Output Format Rules

• Always show: symbol, price, change %, signal, top 3 headlines
• If multiple stocks: group by signal (BUY first, then HOLD, then SELL)
• Use emoji prefix: 🟢 BUY / 🟡 HOLD / 🔴 SELL
• Append ⚠️ ALERT to any symbol that crossed a threshold
• End every report with a disclaimer: "This is not financial advice. Data sourced from Google Finance."

---

Limitations & Notes

• Google Finance does not provide real-time Level 2 data; prices may be 15 min delayed for some exchanges.
• This skill cannot execute trades. Recommendations are informational only.
• For non-US stocks, use the exchange suffix (e.g. 0700.HK, 7203.T, BABA.N). See data-sources.md for the full mapping.
• If Google Finance blocks the browser session, fall back to Yahoo Finance scraping as described in data-sources.md.

---

External Endpoints

This skill makes outbound requests to the following public URLs only:

No user data, credentials, or personal information is sent to any external endpoint.

---

Security & Privacy

• No credentials required. All data is fetched from public, unauthenticated endpoints.
• Local state only. The watchlist and price snapshots are stored exclusively at ~/.openclaw/workspace/stock-tracker-state.json on your machine. Nothing is sent to remote servers.
• No code execution from web content. The browser is used read-only to scrape DOM text. No scripts from Google Finance are executed by this skill.
• parse-stock.py is sandboxed. It reads/writes only the state file at the path above. It does not access environment variables, network sockets, or other files.
• Buy/sell signals are heuristic only. No financial data or decisions are transmitted anywhere. All analysis runs locally.