restore and rebuild all stock query scripts

Recover 7 existing scripts from remote repo and rebuild 5 missing ones: - twse-margin-trading.py (融資融券) - twse-day-trading.py (當沖交易) - twse-foreign-holdings.py (外資持股) - yahoo-dividend-browserless.py (股利資訊) - yahoo-broker-trading-browserless.py (主力券商進出) Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-10 10:07:31 +08:00
commit 7d72aed984
16 changed files with 2153 additions and 0 deletions
--- a/.gitignore
+++ b/.gitignore
@@ -0,0 +1,19 @@
+# Python
+__pycache__/
+*.py[cod]
+*$py.class
+*.so
+.Python
+*.egg-info/
+dist/
+build/
+
+# IDE
+.vscode/
+.idea/
+*.swp
+*.swo
+
+# OS
+.DS_Store
+Thumbs.db
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -0,0 +1,107 @@
+# CLAUDE.md
+
+This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
+
+## Overview
+
+This is a stock query skill for Taiwan stock market data. Python scripts in `scripts/` fetch data from Yahoo Finance Taiwan using a Browserless headless browser service, then output concise Chinese summaries.
+
+## Environment Variables
+
+All `*-browserless.py` scripts require these environment variables (loaded from `.env`):
+- `BROWSERLESS_ENDPOINT` - Browserless service URL (e.g., `http://192.168.42.124:13000`)
+- `BROWSERLESS_TOKEN` - Authentication token for Browserless
+
+The `.env` file is searched first in current directory, then at workspace root (3 levels up from scripts).
+
+## Architecture
+
+Scripts fall into two categories:
+
+### 1. Browserless Scripts (Dynamic Content Scraping)
+Use Browserless to fetch Yahoo pages that require JavaScript rendering:
+
+| Script | Purpose |
+|--------|---------|
+| `yahoo-quote-browserless.py` | Individual stock quote (price, OHLC, volume, bid/ask) |
+| `yahoo-institutional-browserless.py` | Three major institutional investors (foreign, trust, dealer) |
+| `yahoo-technical-browserless.py` | Technical indicators (RSI14, ATR14, KD, Bollinger Bands, 5-day K-line) |
+| `yahoo-market-browserless.py` | TAIEX weighted index (^TWII) market summary |
+| `yahoo-dividend-browserless.py` | Dividend history (cash/stock dividend, ex-date, yield) |
+| `yahoo-broker-trading-browserless.py` | Top broker buy/sell rankings |
+
+### 2. Direct API Scripts (No Browserless Required)
+Use Yahoo Finance Chart API directly:
+
+| Script | Purpose |
+|--------|---------|
+| `yahoo-etf-holdings.py` | ETF holdings (top 10, industry distribution, asset allocation) |
+| `yahoo-ma-signal.py` | Moving averages (MA5/10/20/60) and 20-day bias |
+| `yahoo-macd-signal.py` | MACD indicator (DIF, DEA, histogram) |
+
+### 3. TWSE Public API Scripts (No Browserless Required)
+Use TWSE open data API directly:
+
+| Script | Purpose |
+|--------|---------|
+| `twse-margin-trading.py` | Margin trading balance (融資融券餘額) |
+| `twse-day-trading.py` | Day trading statistics (當沖交易統計) |
+| `twse-foreign-holdings.py` | Foreign investor holdings ratio (外資持股比例) |
+
+## Common Commands
+
+### Query individual stock (full analysis)
+```bash
+# Three scripts are typically run together for complete picture
+python3 skills/stocks-query/scripts/yahoo-institutional-browserless.py 2330
+python3 skills/stocks-query/scripts/yahoo-quote-browserless.py 2330
+python3 skills/stocks-query/scripts/yahoo-technical-browserless.py 2330
+```
+
+### Query ETF (with holdings)
+```bash
+python3 skills/stocks-query/scripts/yahoo-quote-browserless.py 0050
+python3 skills/stocks-query/scripts/yahoo-etf-holdings.py 0050
+```
+
+### Query market index
+```bash
+python3 skills/stocks-query/scripts/yahoo-market-browserless.py
+```
+
+### MA/MACD signals (direct API, no Browserless)
+```bash
+python3 skills/stocks-query/scripts/yahoo-ma-signal.py 2330
+python3 skills/stocks-query/scripts/yahoo-macd-signal.py 2330
+```
+
+### Dividend / Broker trading (Browserless)
+```bash
+python3 skills/stocks-query/scripts/yahoo-dividend-browserless.py 2330
+python3 skills/stocks-query/scripts/yahoo-broker-trading-browserless.py 2330
+```
+
+### TWSE data (margin, day-trading, foreign holdings)
+```bash
+python3 skills/stocks-query/scripts/twse-margin-trading.py 2330
+python3 skills/stocks-query/scripts/twse-day-trading.py 2330
+python3 skills/stocks-query/scripts/twse-foreign-holdings.py 2330
+```
+
+## Script Patterns
+
+- All scripts accept stock symbol as first command-line argument (default provided)
+- Browserless scripts use `urllib` or `subprocess` with `curl` to POST to Browserless `/content` endpoint
+- HTML parsing uses regex patterns (not BeautifulSoup) to minimize dependencies
+- Output format is Chinese text summaries with bullet points
+- Error handling exits with code 1 and prints error message to stderr
+
+## Stock Number Format
+
+- Taiwan stocks: 4-digit number (e.g., `2330`, `3481`)
+- ETFs: Typically start with `00` (e.g., `0050`, `00922`)
+- Yahoo suffix: `.TW` is appended automatically in URLs
+
+## Skill Integration
+
+This directory is registered as a skill (see `SKILL.md`). When users input a stock number or ask about "大盤" (market), the scripts should be executed in sequence and results synthesized into a concise Chinese summary with plain-language interpretation of technical indicators.
--- a/SKILL.md
+++ b/SKILL.md
@@ -0,0 +1,84 @@
+---
+name: stocks-query
+description: 查詢台股個股與大盤（Yahoo 資料）並整理重點摘要。當使用者輸入股票代號（例如 2330、0050、3481）或詢問「大盤」時使用。流程：個股依序執行法人、報價、技術分析三支 Python 腳本；大盤執行大盤腳本；最後輸出精簡中文摘要，並檢查是否有本地持股庫存紀錄。
+---
+
+# stocks-query
+
+使用 `scripts/stocks/` 內的腳本查詢台股資料，並回覆精簡摘要。
+
+## 執行流程
+
+### A) 使用者只輸入股票代號（例：2330、0050、3481）
+
+依序執行（盡量平行呼叫以加快速度）：
+
+```bash
+python3 skills/stocks-query/scripts/yahoo-institutional-browserless.py <代號>
+python3 skills/stocks-query/scripts/yahoo-quote-browserless.py <代號>
+python3 skills/stocks-query/scripts/yahoo-technical-browserless.py <代號>
+python3 skills/stocks-query/scripts/twse-margin-trading.py <代號>
+python3 skills/stocks-query/scripts/yahoo-dividend-browserless.py <代號>
+python3 skills/stocks-query/scripts/yahoo-market-brief.py
+python3 skills/stocks-query/scripts/yahoo-broker-trading-browserless.py <代號>
+python3 skills/stocks-query/scripts/twse-day-trading.py <代號>
+python3 skills/stocks-query/scripts/twse-foreign-holdings.py <代號>
+```
+
+若標的**不是** ETF（非 `00` 開頭），額外執行：
+
+```bash
+python3 skills/stocks-query/scripts/twse-monthly-revenue.py <代號>
+python3 skills/stocks-query/scripts/twse-pe-pbr.py <代號>
+```
+
+若標的是 ETF（`00` 開頭），改為額外執行：
+
+```bash
+python3 skills/stocks-query/scripts/yahoo-etf-holdings.py <代號>
+```
+
+注意：月營收、本益比 對 ETF 無資料，不要對 ETF 執行。
+
+再檢查本地是否有該代號的持股／庫存紀錄（例如 `stocks/`、`purchases/` 相關檔）。
+
+回覆重點：
+- 現價、漲跌、開高低、昨收、成交量、均價、最佳買賣
+- 三大法人（外資／投信／自營商／合計）
+- 近 5 日 K 線（開高低收）
+- RSI14
+- ATR14
+- KD
+- 布林通道
+- 看均線、量能、乖離、MACD（若腳本有提供資料則納入摘要）
+- 融資融券：融資餘額（增減）、融券餘額（增減）、券資比、融資使用率
+- 股利殖利率：歷年現金股利、殖利率、最近除息日與填息天數
+- 大盤同步：加權指數現值與漲跌（一行簡要）
+- 主力券商：買超／賣超前 15 大券商、主力買賣超合計張數
+- 當沖統計：當沖成交量、佔個股成交比例、大盤當沖佔比
+- 外資持股：外資持有股數與持股比例
+- 非 ETF 額外補：月營收（當月、MoM、YoY、累計）、本益比、股價淨值比
+- ETF 額外補：前十大持股、行業比重、資產分佈
+- 是否查到本地持股紀錄
+- 以上技術面數值後，必須補一段「白話判斷」，用台灣繁中解釋目前偏強／偏弱、是否過熱、波動大不大、短線偏多或偏震盪，避免只丟數字。
+
+### B) 使用者詢問「大盤」
+
+執行：
+
+```bash
+python3 skills/stocks-query/scripts/yahoo-market-browserless.py
+```
+
+回覆重點：
+- 加權指數
+- 漲跌（點數與百分比）
+- 開高低、昨收
+- 成交金額
+- 更新時間
+
+## 回覆風格
+
+- 使用繁體中文（台灣）
+- 簡潔、重點式條列
+- 若某欄位缺值，明確標註「無資料」
--- a/scripts/twse-day-trading.py
+++ b/scripts/twse-day-trading.py
@@ -0,0 +1,131 @@
+#!/usr/bin/env python3
+"""
+查詢 TWSE 個股當日沖銷交易統計
+
+用法：
+  python3 skills/stocks-query/scripts/twse-day-trading.py 2330
+  python3 skills/stocks-query/scripts/twse-day-trading.py 3481
+
+資料來源：證交所 TWSE 公開 API（當日沖銷交易標的及成交量值）
+"""
+
+from __future__ import annotations
+
+import json
+import sys
+import urllib.request
+from datetime import datetime, timedelta
+
+
+def fetch_day_trading(stock_no: str, date_str: str | None = None) -> dict:
+    """從 TWSE 抓當沖資料，自動往前回溯最多 10 天找到有資料的日期。"""
+    if date_str:
+        dates_to_try = [date_str]
+    else:
+        today = datetime.now()
+        dates_to_try = [(today - timedelta(days=i)).strftime("%Y%m%d") for i in range(10)]
+
+    for d in dates_to_try:
+        url = (
+            f"https://www.twse.com.tw/rwd/zh/dayTrading/dayTrading"
+            f"?date={d}&selectType=ALL&response=json"
+        )
+        req = urllib.request.Request(url, headers={"User-Agent": "Mozilla/5.0"})
+        try:
+            with urllib.request.urlopen(req, timeout=30) as resp:
+                data = json.loads(resp.read().decode("utf-8", errors="replace"))
+        except Exception:
+            continue
+
+        if data.get("stat") != "OK":
+            continue
+
+        tables = data.get("tables", [])
+        if not tables:
+            # 有些 API 回傳格式不同，直接看 data 欄位
+            fields = data.get("fields", [])
+            rows = data.get("data", [])
+            if fields and rows:
+                tables = [{"fields": fields, "data": rows}]
+            else:
+                continue
+
+        # 找含「代號」欄位的表
+        target_table = None
+        for tbl in tables:
+            fields = tbl.get("fields", [])
+            field_str = "".join(fields)
+            if "代號" in field_str or "證券代號" in field_str:
+                target_table = tbl
+                break
+
+        if not target_table:
+            # 嘗試用第一個 table
+            if tables:
+                target_table = tables[0]
+            else:
+                continue
+
+        fields = target_table["fields"]
+
+        # 找代號欄位的 index
+        code_idx = None
+        for i, f in enumerate(fields):
+            if "代號" in f:
+                code_idx = i
+                break
+
+        if code_idx is None:
+            continue
+
+        for row in target_table.get("data", []):
+            code = row[code_idx].strip()
+            if code == stock_no:
+                result = {"date": d}
+                for i, field in enumerate(fields):
+                    result[field.strip()] = row[i].strip().replace(",", "") if i < len(row) else "-"
+                return result
+
+    raise RuntimeError(f"找不到 {stock_no} 的當沖資料（已回溯 10 天）")
+
+
+def print_summary(stock_no: str, d: dict) -> None:
+    date_fmt = f"{d['date'][:4]}/{d['date'][4:6]}/{d['date'][6:]}"
+    name = d.get("證券名稱", d.get("名稱", stock_no))
+    print(f"{name}（{stock_no}）｜當沖交易摘要")
+    print(f"• 日期：{date_fmt}")
+
+    # 當沖買賣量
+    day_buy = d.get("當日沖銷交易買進成交股數", d.get("買進成交股數", "-"))
+    day_sell = d.get("當日沖銷交易賣出成交股數", d.get("賣出成交股數", "-"))
+    day_amount = d.get("當日沖銷交易成交金額", d.get("成交金額", "-"))
+
+    print(f"• 當沖買進股數：{day_buy}")
+    print(f"• 當沖賣出股數：{day_sell}")
+
+    if day_amount != "-":
+        print(f"• 當沖成交金額：{day_amount}")
+
+    # 當沖比例
+    ratio = d.get("當日沖銷交易佔該日成交量比率", d.get("佔成交量比率", "-"))
+    if ratio == "-":
+        # 嘗試自己算
+        try:
+            total_vol = float(d.get("成交股數", "0"))
+            day_vol = float(day_buy) if day_buy != "-" else 0
+            if total_vol > 0 and day_vol > 0:
+                ratio = f"{(day_vol / total_vol) * 100:.2f}%"
+        except (ValueError, ZeroDivisionError):
+            pass
+
+    print(f"• 當沖佔比：{ratio}")
+
+
+if __name__ == "__main__":
+    stock_no = sys.argv[1] if len(sys.argv) > 1 else "2330"
+    try:
+        data = fetch_day_trading(stock_no)
+        print_summary(stock_no, data)
+    except Exception as e:
+        print(f"取得失敗：{e}", file=sys.stderr)
+        sys.exit(1)
--- a/scripts/twse-foreign-holdings.py
+++ b/scripts/twse-foreign-holdings.py
@@ -0,0 +1,128 @@
+#!/usr/bin/env python3
+"""
+查詢 TWSE 外資持股比例統計
+
+用法：
+  python3 skills/stocks-query/scripts/twse-foreign-holdings.py 2330
+  python3 skills/stocks-query/scripts/twse-foreign-holdings.py 3481
+
+資料來源：證交所 TWSE 公開 API（外資及陸資持股統計）
+"""
+
+from __future__ import annotations
+
+import json
+import sys
+import urllib.request
+from datetime import datetime, timedelta
+
+
+def fetch_foreign_holdings(stock_no: str, date_str: str | None = None) -> dict:
+    """從 TWSE 抓外資持股資料，自動往前回溯最多 10 天找到有資料的日期。"""
+    if date_str:
+        dates_to_try = [date_str]
+    else:
+        today = datetime.now()
+        dates_to_try = [(today - timedelta(days=i)).strftime("%Y%m%d") for i in range(10)]
+
+    for d in dates_to_try:
+        url = (
+            f"https://www.twse.com.tw/rwd/zh/fund/MI_QFIIS"
+            f"?date={d}&selectType=ALL&response=json"
+        )
+        req = urllib.request.Request(url, headers={"User-Agent": "Mozilla/5.0"})
+        try:
+            with urllib.request.urlopen(req, timeout=30) as resp:
+                data = json.loads(resp.read().decode("utf-8", errors="replace"))
+        except Exception:
+            continue
+
+        if data.get("stat") != "OK":
+            continue
+
+        tables = data.get("tables", [])
+        if not tables:
+            fields = data.get("fields", [])
+            rows = data.get("data", [])
+            if fields and rows:
+                tables = [{"fields": fields, "data": rows}]
+            else:
+                continue
+
+        # 找含「證券代號」或「代號」欄位的表
+        target_table = None
+        for tbl in tables:
+            fields = tbl.get("fields", [])
+            field_str = "".join(fields)
+            if "代號" in field_str:
+                target_table = tbl
+                break
+
+        if not target_table:
+            if tables:
+                target_table = tables[0]
+            else:
+                continue
+
+        fields = target_table["fields"]
+
+        # 找代號欄位 index
+        code_idx = None
+        for i, f in enumerate(fields):
+            if "代號" in f:
+                code_idx = i
+                break
+
+        if code_idx is None:
+            continue
+
+        for row in target_table.get("data", []):
+            code = row[code_idx].strip()
+            if code == stock_no:
+                result = {"date": d}
+                for i, field in enumerate(fields):
+                    result[field.strip()] = row[i].strip().replace(",", "") if i < len(row) else "-"
+                return result
+
+    raise RuntimeError(f"找不到 {stock_no} 的外資持股資料（已回溯 10 天）")
+
+
+def print_summary(stock_no: str, d: dict) -> None:
+    date_fmt = f"{d['date'][:4]}/{d['date'][4:6]}/{d['date'][6:]}"
+    name = d.get("證券名稱", d.get("名稱", stock_no))
+    print(f"{name}（{stock_no}）｜外資持股摘要")
+    print(f"• 日期：{date_fmt}")
+
+    # 外資持股
+    shares = d.get("外資及陸資持股股數", d.get("外陸資持股股數", "-"))
+    pct = d.get("外資及陸資持股比率", d.get("外陸資持股比率", d.get("持股比率", "-")))
+    available = d.get("外資及陸資尚可投資股數", d.get("尚可投資股數", "-"))
+    upper_limit = d.get("外資及陸資共用法令投資上限比率", d.get("投資上限比率", "-"))
+    total_issued = d.get("發行股數", "-")
+
+    # 外資買賣超相關
+    foreign_buy = d.get("外資及陸資買進股數", d.get("買進股數", "-"))
+    foreign_sell = d.get("外資及陸資賣出股數", d.get("賣出股數", "-"))
+
+    print(f"• 外資持股股數：{shares}")
+    print(f"• 外資持股比率：{pct}%") if pct != "-" and "%" not in str(pct) else print(f"• 外資持股比率：{pct}")
+    if total_issued != "-":
+        print(f"• 發行股數：{total_issued}")
+    if available != "-":
+        print(f"• 尚可投資股數：{available}")
+    if upper_limit != "-":
+        print(f"• 投資上限比率：{upper_limit}")
+    if foreign_buy != "-":
+        print(f"• 外資買進股數：{foreign_buy}")
+    if foreign_sell != "-":
+        print(f"• 外資賣出股數：{foreign_sell}")
+
+
+if __name__ == "__main__":
+    stock_no = sys.argv[1] if len(sys.argv) > 1 else "2330"
+    try:
+        data = fetch_foreign_holdings(stock_no)
+        print_summary(stock_no, data)
+    except Exception as e:
+        print(f"取得失敗：{e}", file=sys.stderr)
+        sys.exit(1)
--- a/scripts/twse-margin-trading.py
+++ b/scripts/twse-margin-trading.py
@@ -0,0 +1,120 @@
+#!/usr/bin/env python3
+"""
+查詢 TWSE 個股融資融券餘額（信用交易統計）
+
+用法：
+  python3 skills/stocks-query/scripts/twse-margin-trading.py 2330
+  python3 skills/stocks-query/scripts/twse-margin-trading.py 3481
+
+資料來源：證交所 TWSE 公開 API
+"""
+
+from __future__ import annotations
+
+import json
+import sys
+import urllib.request
+from datetime import datetime, timedelta
+
+
+def fetch_margin(stock_no: str, date_str: str | None = None) -> dict:
+    """從 TWSE 抓融資融券資料，自動往前回溯最多 10 天找到有資料的日期。"""
+    if date_str:
+        dates_to_try = [date_str]
+    else:
+        today = datetime.now()
+        dates_to_try = [(today - timedelta(days=i)).strftime("%Y%m%d") for i in range(10)]
+
+    for d in dates_to_try:
+        url = (
+            f"https://www.twse.com.tw/rwd/zh/marginTrading/MI_MARGN"
+            f"?date={d}&selectType=ALL&response=json"
+        )
+        req = urllib.request.Request(url, headers={"User-Agent": "Mozilla/5.0"})
+        try:
+            with urllib.request.urlopen(req, timeout=30) as resp:
+                data = json.loads(resp.read().decode("utf-8", errors="replace"))
+        except Exception:
+            continue
+
+        if data.get("stat") != "OK":
+            continue
+
+        tables = data.get("tables", [])
+        if not tables:
+            continue
+
+        # 找到個股那張表（title 含「融資融券」或欄位含「股票代號」）
+        target_table = None
+        for tbl in tables:
+            fields = tbl.get("fields", [])
+            if "股票代號" in fields:
+                target_table = tbl
+                break
+
+        if not target_table:
+            continue
+
+        fields = target_table["fields"]
+        for row in target_table.get("data", []):
+            code = row[fields.index("股票代號")].strip()
+            if code == stock_no:
+                result = {"date": d}
+                for i, field in enumerate(fields):
+                    result[field.strip()] = row[i].strip().replace(",", "")
+                return result
+
+    raise RuntimeError(f"找不到 {stock_no} 的融資融券資料（已回溯 10 天）")
+
+
+def print_summary(stock_no: str, d: dict) -> None:
+    date_fmt = f"{d['date'][:4]}/{d['date'][4:6]}/{d['date'][6:]}"
+    name = d.get("股票名稱", stock_no)
+    print(f"{name}（{stock_no}）｜融資融券摘要")
+    print(f"• 日期：{date_fmt}")
+
+    # 融資
+    margin_buy = d.get("融資買進", "-")
+    margin_sell = d.get("融資賣出", "-")
+    margin_cash = d.get("融資現償", "-")
+    margin_balance = d.get("融資餘額", d.get("融資今日餘額", "-"))
+    margin_limit = d.get("融資限額", "-")
+    print(f"• 融資買進：{margin_buy}")
+    print(f"• 融資賣出：{margin_sell}")
+    print(f"• 融資現償：{margin_cash}")
+    print(f"• 融資餘額：{margin_balance}")
+    if margin_limit != "-":
+        print(f"• 融資限額：{margin_limit}")
+
+    # 融券
+    short_sell = d.get("融券賣出", "-")
+    short_buy = d.get("融券買進", "-")
+    short_cash = d.get("融券現償", "-")
+    short_balance = d.get("融券餘額", d.get("融券今日餘額", "-"))
+    short_limit = d.get("融券限額", "-")
+    print(f"• 融券賣出：{short_sell}")
+    print(f"• 融券買進：{short_buy}")
+    print(f"• 融券現償：{short_cash}")
+    print(f"• 融券餘額：{short_balance}")
+    if short_limit != "-":
+        print(f"• 融券限額：{short_limit}")
+
+    # 券資比
+    try:
+        mb = float(margin_balance) if margin_balance != "-" else 0
+        sb = float(short_balance) if short_balance != "-" else 0
+        if mb > 0:
+            ratio = (sb / mb) * 100
+            print(f"• 券資比：{ratio:.2f}%")
+    except (ValueError, ZeroDivisionError):
+        pass
+
+
+if __name__ == "__main__":
+    stock_no = sys.argv[1] if len(sys.argv) > 1 else "2330"
+    try:
+        data = fetch_margin(stock_no)
+        print_summary(stock_no, data)
+    except Exception as e:
+        print(f"取得失敗：{e}", file=sys.stderr)
+        sys.exit(1)
--- a/scripts/yahoo-broker-trading-browserless.py
+++ b/scripts/yahoo-broker-trading-browserless.py
@@ -0,0 +1,215 @@
+#!/usr/bin/env python3
+"""
+用 Browserless 抓 Yahoo 主力券商進出頁，輸出精簡摘要
+
+用法：
+  python3 skills/stocks-query/scripts/yahoo-broker-trading-browserless.py 2330
+  python3 skills/stocks-query/scripts/yahoo-broker-trading-browserless.py 3481
+
+環境變數（可覆蓋）：
+  BROWSERLESS_ENDPOINT=http://192.168.42.124:13000
+  BROWSERLESS_TOKEN=6R0W53R135510
+"""
+
+import json
+import os
+import re
+import sys
+import urllib.error
+import urllib.request
+from pathlib import Path
+
+
+def load_dotenv() -> None:
+    """載入 .env（優先目前目錄，其次 workspace 根目錄）。"""
+    candidates = [
+        Path.cwd() / ".env",
+        Path(__file__).resolve().parents[3] / ".env",
+    ]
+    for env_path in candidates:
+        if not env_path.exists():
+            continue
+        for raw in env_path.read_text(encoding="utf-8").splitlines():
+            line = raw.strip()
+            if not line or line.startswith("#") or "=" not in line:
+                continue
+            k, v = line.split("=", 1)
+            k = k.strip()
+            v = v.strip().strip('"').strip("'")
+            if k and k not in os.environ:
+                os.environ[k] = v
+        break
+
+
+load_dotenv()
+
+ENDPOINT = os.getenv("BROWSERLESS_ENDPOINT")
+TOKEN = os.getenv("BROWSERLESS_TOKEN")
+if not ENDPOINT or not TOKEN:
+    raise RuntimeError("缺少必要環境變數：BROWSERLESS_ENDPOINT / BROWSERLESS_TOKEN")
+STOCK_NO = sys.argv[1] if len(sys.argv) > 1 else "2330"
+
+
+def fetch_html(stock_no: str) -> str:
+    url = f"https://tw.stock.yahoo.com/quote/{stock_no}.TW/broker-trading"
+    api = f"{ENDPOINT}/content?token={TOKEN}"
+    payload = {
+        "url": url,
+        "gotoOptions": {"waitUntil": "domcontentloaded", "timeout": 60000},
+    }
+
+    req = urllib.request.Request(
+        api,
+        data=json.dumps(payload).encode("utf-8"),
+        headers={"Content-Type": "application/json"},
+        method="POST",
+    )
+
+    try:
+        with urllib.request.urlopen(req, timeout=90) as resp:
+            text = resp.read().decode("utf-8", errors="replace")
+            status = resp.getcode()
+    except urllib.error.HTTPError as e:
+        body = e.read().decode("utf-8", errors="replace") if hasattr(e, "read") else str(e)
+        raise RuntimeError(f"Browserless HTTP {e.code}: {body[:200]}") from e
+
+    if status != 200:
+        raise RuntimeError(f"Browserless HTTP {status}: {text[:200]}")
+    if "TimeoutError" in text:
+        raise RuntimeError("Browserless 導航逾時（TimeoutError）")
+    return text
+
+
+def pick(html: str, pattern: str, default: str = "-") -> str:
+    m = re.search(pattern, html, flags=re.S)
+    return m.group(1) if m else default
+
+
+def parse_brokers(html: str, stock_no: str) -> dict:
+    name = pick(html, r'property="og:title" content="([^"(]+)\([0-9]+\.TW\)', "").strip()
+    if not name:
+        name = pick(html, r"<title>([^<(]+)\([0-9]+\.TW\)", "-").strip()
+
+    # 嘗試從嵌入 JSON 中抓券商進出資料
+    buy_brokers = []
+    sell_brokers = []
+
+    # 方法1：從 brokerBuySellData / topBrokers JSON 結構抓取
+    json_match = re.search(r'"brokerBuySellData"\s*:\s*(\{[\s\S]*?\})\s*[,}]', html)
+    if json_match:
+        try:
+            raw = json.loads(json_match.group(1))
+            for item in raw.get("buy", raw.get("buyBrokers", []))[:10]:
+                buy_brokers.append({
+                    "name": item.get("brokerName", item.get("name", "-")),
+                    "buy": str(item.get("buy", item.get("buyVolume", 0))),
+                    "sell": str(item.get("sell", item.get("sellVolume", 0))),
+                    "diff": str(item.get("diff", item.get("net", 0))),
+                })
+            for item in raw.get("sell", raw.get("sellBrokers", []))[:10]:
+                sell_brokers.append({
+                    "name": item.get("brokerName", item.get("name", "-")),
+                    "buy": str(item.get("buy", item.get("buyVolume", 0))),
+                    "sell": str(item.get("sell", item.get("sellVolume", 0))),
+                    "diff": str(item.get("diff", item.get("net", 0))),
+                })
+        except (json.JSONDecodeError, KeyError):
+            pass
+
+    # 方法2：從 HTML table 結構抓取
+    if not buy_brokers and not sell_brokers:
+        # 買超券商區塊
+        buy_section = ""
+        sell_section = ""
+        buy_start = html.find("買超券商")
+        sell_start = html.find("賣超券商")
+
+        if buy_start != -1 and sell_start != -1:
+            buy_section = html[buy_start:sell_start]
+            sell_section = html[sell_start:sell_start + 50000]
+        elif buy_start != -1:
+            buy_section = html[buy_start:buy_start + 50000]
+
+        broker_row_pat = re.compile(
+            r'<li[^>]*>.*?'
+            r'(?:券商|broker)[^<]*</[^>]+>.*?'
+            r'([^<]{2,20})</(?:span|div|td)>'
+            r'.*?([\d,]+).*?([\d,]+).*?([+-]?[\d,]+)',
+            re.S,
+        )
+
+        # 更通用的抓法：連續出現的券商名+數字 pattern
+        simple_pat = re.compile(
+            r'>([^\d<>]{2,12}(?:證券|期貨|投顧|銀行|金控)?[^<>]{0,6})</(?:span|div|td|a)>'
+            r'[^<]*<[^>]+>([\d,]+)</[^>]+>'
+            r'[^<]*<[^>]+>([\d,]+)</[^>]+>'
+            r'[^<]*<[^>]+>([+-]?[\d,]+)</[^>]+>',
+            re.S,
+        )
+
+        for section, target_list in [(buy_section, buy_brokers), (sell_section, sell_brokers)]:
+            for m in simple_pat.finditer(section):
+                broker_name = m.group(1).strip()
+                if len(broker_name) < 2 or broker_name.isdigit():
+                    continue
+                target_list.append({
+                    "name": broker_name,
+                    "buy": m.group(2).replace(",", ""),
+                    "sell": m.group(3).replace(",", ""),
+                    "diff": m.group(4).replace(",", ""),
+                })
+                if len(target_list) >= 10:
+                    break
+
+    # 日期
+    date = pick(html, r"資料日期[：:]\s*([0-9/\-]+)")
+    if date == "-":
+        date = pick(html, r'"formattedDate"\s*:\s*"([^"]+)"')
+
+    return {
+        "name": name,
+        "stock_no": stock_no,
+        "date": date,
+        "buy_brokers": buy_brokers[:10],
+        "sell_brokers": sell_brokers[:10],
+    }
+
+
+def sign(v: str) -> str:
+    if not v or v == "-":
+        return "-"
+    try:
+        n = int(v.replace(",", ""))
+        return f"+{n:,}" if n >= 0 else f"{n:,}"
+    except ValueError:
+        return v
+
+
+def print_summary(d: dict) -> None:
+    print(f"{d['name']}（{d['stock_no']}）｜主力券商進出")
+    if d["date"] != "-":
+        print(f"• 日期：{d['date']}")
+
+    if d["buy_brokers"]:
+        print("• 買超前幾大券商：")
+        for b in d["buy_brokers"][:5]:
+            print(f"  - {b['name']}：買 {b['buy']}／賣 {b['sell']}／差 {sign(b['diff'])} 張")
+    else:
+        print("• 買超券商：無資料")
+
+    if d["sell_brokers"]:
+        print("• 賣超前幾大券商：")
+        for b in d["sell_brokers"][:5]:
+            print(f"  - {b['name']}：買 {b['buy']}／賣 {b['sell']}／差 {sign(b['diff'])} 張")
+    else:
+        print("• 賣超券商：無資料")
+
+
+if __name__ == "__main__":
+    try:
+        html = fetch_html(STOCK_NO)
+        data = parse_brokers(html, STOCK_NO)
+        print_summary(data)
+    except Exception as e:
+        print(f"取得失敗：{e}", file=sys.stderr)
+        sys.exit(1)
--- a/scripts/yahoo-dividend-browserless.py
+++ b/scripts/yahoo-dividend-browserless.py
@@ -0,0 +1,190 @@
+#!/usr/bin/env python3
+"""
+用 Browserless 抓 Yahoo 股利政策頁，輸出精簡摘要
+
+用法：
+  python3 skills/stocks-query/scripts/yahoo-dividend-browserless.py 2330
+  python3 skills/stocks-query/scripts/yahoo-dividend-browserless.py 0050
+
+環境變數（可覆蓋）：
+  BROWSERLESS_ENDPOINT=http://192.168.42.124:13000
+  BROWSERLESS_TOKEN=6R0W53R135510
+"""
+
+import json
+import os
+import re
+import sys
+import urllib.error
+import urllib.request
+from pathlib import Path
+
+
+def load_dotenv() -> None:
+    """載入 .env（優先目前目錄，其次 workspace 根目錄）。"""
+    candidates = [
+        Path.cwd() / ".env",
+        Path(__file__).resolve().parents[3] / ".env",
+    ]
+    for env_path in candidates:
+        if not env_path.exists():
+            continue
+        for raw in env_path.read_text(encoding="utf-8").splitlines():
+            line = raw.strip()
+            if not line or line.startswith("#") or "=" not in line:
+                continue
+            k, v = line.split("=", 1)
+            k = k.strip()
+            v = v.strip().strip('"').strip("'")
+            if k and k not in os.environ:
+                os.environ[k] = v
+        break
+
+
+load_dotenv()
+
+ENDPOINT = os.getenv("BROWSERLESS_ENDPOINT")
+TOKEN = os.getenv("BROWSERLESS_TOKEN")
+if not ENDPOINT or not TOKEN:
+    raise RuntimeError("缺少必要環境變數：BROWSERLESS_ENDPOINT / BROWSERLESS_TOKEN")
+STOCK_NO = sys.argv[1] if len(sys.argv) > 1 else "2330"
+
+
+def fetch_html(stock_no: str) -> str:
+    url = f"https://tw.stock.yahoo.com/quote/{stock_no}.TW/dividend"
+    api = f"{ENDPOINT}/content?token={TOKEN}"
+    payload = {
+        "url": url,
+        "gotoOptions": {"waitUntil": "domcontentloaded", "timeout": 60000},
+    }
+
+    req = urllib.request.Request(
+        api,
+        data=json.dumps(payload).encode("utf-8"),
+        headers={"Content-Type": "application/json"},
+        method="POST",
+    )
+
+    try:
+        with urllib.request.urlopen(req, timeout=90) as resp:
+            text = resp.read().decode("utf-8", errors="replace")
+            status = resp.getcode()
+    except urllib.error.HTTPError as e:
+        body = e.read().decode("utf-8", errors="replace") if hasattr(e, "read") else str(e)
+        raise RuntimeError(f"Browserless HTTP {e.code}: {body[:200]}") from e
+
+    if status != 200:
+        raise RuntimeError(f"Browserless HTTP {status}: {text[:200]}")
+    if "TimeoutError" in text:
+        raise RuntimeError("Browserless 導航逾時（TimeoutError）")
+    return text
+
+
+def pick(html: str, pattern: str, default: str = "-") -> str:
+    m = re.search(pattern, html, flags=re.S)
+    return m.group(1) if m else default
+
+
+def parse_dividend(html: str, stock_no: str) -> dict:
+    name = pick(html, r'property="og:title" content="([^"(]+)\([0-9]+\.TW\)', "").strip()
+    if not name:
+        name = pick(html, r"<title>([^<(]+)\([0-9]+\.TW\)", "-").strip()
+
+    # 殖利率
+    dividend_yield = pick(html, r">現金殖利率</span><span[^>]*>([\d.]+%?)")
+
+    # 年度股利統計（從 JSON-like 嵌入或 table 抓取）
+    # 嘗試從頁面的 script data 抓股利歷史
+    rows = []
+
+    # 方法1：抓 dividendData JSON
+    json_match = re.search(r'"dividendData"\s*:\s*(\[[\s\S]*?\])\s*[,}]', html)
+    if json_match:
+        try:
+            raw_data = json.loads(json_match.group(1))
+            for item in raw_data[:5]:  # 最近 5 年
+                year = item.get("year", item.get("formattedYear", "-"))
+                cash = item.get("cashDividend", item.get("cash", 0))
+                stock = item.get("stockDividend", item.get("stock", 0))
+                total = item.get("totalDividend", 0)
+                if total == 0:
+                    try:
+                        total = float(cash) + float(stock)
+                    except (ValueError, TypeError):
+                        total = "-"
+                rows.append({
+                    "year": str(year),
+                    "cash": str(cash),
+                    "stock": str(stock),
+                    "total": str(total),
+                })
+        except (json.JSONDecodeError, KeyError):
+            pass
+
+    # 方法2：用 regex 抓 table rows
+    if not rows:
+        # 常見結構：年度、現金股利、股票股利、合計
+        row_pattern = re.compile(
+            r'<tr[^>]*>.*?<td[^>]*>(1\d{2,3})</td>'
+            r'.*?<td[^>]*>([\d.]+)</td>'
+            r'.*?<td[^>]*>([\d.]+)</td>'
+            r'.*?<td[^>]*>([\d.]+)</td>',
+            re.S,
+        )
+        for m in row_pattern.finditer(html):
+            rows.append({
+                "year": m.group(1),
+                "cash": m.group(2),
+                "stock": m.group(3),
+                "total": m.group(4),
+            })
+            if len(rows) >= 5:
+                break
+
+    # 方法3：抓除息日相關
+    ex_date = pick(html, r">除息日</span><span[^>]*>([^<]+)")
+    pay_date = pick(html, r">發放日</span><span[^>]*>([^<]+)")
+    cash_div = pick(html, r">現金股利</span><span[^>]*>([\d.]+)")
+    stock_div = pick(html, r">股票股利</span><span[^>]*>([\d.]+)")
+
+    return {
+        "name": name,
+        "stock_no": stock_no,
+        "dividend_yield": dividend_yield,
+        "cash_div": cash_div,
+        "stock_div": stock_div,
+        "ex_date": ex_date,
+        "pay_date": pay_date,
+        "history": rows[:5],
+    }
+
+
+def print_summary(d: dict) -> None:
+    print(f"{d['name']}（{d['stock_no']}）｜股利摘要")
+    if d["dividend_yield"] != "-":
+        print(f"• 現金殖利率：{d['dividend_yield']}")
+    if d["cash_div"] != "-":
+        print(f"• 現金股利：{d['cash_div']}")
+    if d["stock_div"] != "-":
+        print(f"• 股票股利：{d['stock_div']}")
+    if d["ex_date"] != "-":
+        print(f"• 除息日：{d['ex_date']}")
+    if d["pay_date"] != "-":
+        print(f"• 發放日：{d['pay_date']}")
+
+    if d["history"]:
+        print("• 近年股利：")
+        for r in d["history"]:
+            print(f"  - {r['year']}年：現金 {r['cash']}／股票 {r['stock']}／合計 {r['total']}")
+    else:
+        print("• 近年股利：無資料")
+
+
+if __name__ == "__main__":
+    try:
+        html = fetch_html(STOCK_NO)
+        data = parse_dividend(html, STOCK_NO)
+        print_summary(data)
+    except Exception as e:
+        print(f"取得失敗：{e}", file=sys.stderr)
+        sys.exit(1)
--- a/scripts/yahoo-etf-holdings.py
+++ b/scripts/yahoo-etf-holdings.py
@@ -0,0 +1,164 @@
+#!/usr/bin/env python3
+"""抓 Yahoo 台股 ETF 持股分析頁的前十大持股 / 行業比重 / 資產分佈。"""
+
+from __future__ import annotations
+
+import argparse
+import csv
+import json
+import re
+import sys
+import urllib.request
+from html import unescape
+
+USER_AGENT = "Mozilla/5.0 (OpenClaw; yahoo-etf-holdings)"
+BASE_URL = "https://tw.stock.yahoo.com/quote/{symbol}.TW/holding"
+
+
+def fetch(url: str) -> str:
+    req = urllib.request.Request(url, headers={"User-Agent": USER_AGENT})
+    with urllib.request.urlopen(req, timeout=30) as response:
+        return response.read().decode("utf-8", "ignore")
+
+
+def strip_tags(text: str) -> str:
+    return unescape(re.sub(r"<[^>]+>", "", text)).strip()
+
+
+def extract_date(block: str) -> str:
+    m = re.search(r'<time[^>]*datatime="([0-9]{4}/[0-9]{2}/[0-9]{2})"', block)
+    return m.group(1) if m else ""
+
+
+def extract_block(html: str, start_marker: str, end_marker: str | None = None) -> str:
+    start = html.find(start_marker)
+    if start == -1:
+        return ""
+    sub = html[start:]
+    if end_marker:
+        end = sub.find(end_marker)
+        if end != -1:
+            sub = sub[:end]
+    return sub
+
+
+def parse_rank_list(block: str, value_class: str, limit: int | None = None) -> list[dict[str, str]]:
+    pattern = re.compile(
+        rf'">(\d+)\.</div>([^<]+)</div><div class="{re.escape(value_class)}">([0-9]+\.[0-9]+%)</div>',
+        re.S,
+    )
+    items = []
+    for rank, name, pct in pattern.findall(block):
+        items.append({"排名": rank, "名稱": strip_tags(name), "占比": pct})
+        if limit and len(items) >= limit:
+            break
+    return items
+
+
+def parse_industries(block: str) -> list[dict[str, str]]:
+    pattern = re.compile(
+        r'<div class="D\(f\) Ai\(c\)">(?:<div[^>]*></div>)?([^<]+)</div><div class="Fx\(n\) Fw\(b\) Pstart\(12px\)">([0-9]+\.[0-9]+%)</div>',
+        re.S,
+    )
+    items = []
+    for idx, (name, pct) in enumerate(pattern.findall(block), start=1):
+        items.append({"排名": str(idx), "名稱": strip_tags(name), "占比": pct})
+    return items
+
+
+def parse_holding_page(symbol: str) -> dict:
+    url = BASE_URL.format(symbol=symbol)
+    html = fetch(url)
+
+    asset_block = extract_block(html, '資產分佈</h2>', '前十大持股</h2>')
+    top_block = extract_block(html, '前十大持股</h2>', '網友也在看')
+    industry_block = extract_block(html, '行業比重</h2>', '持股明細</h2>')
+
+    assets = parse_rank_list(asset_block, 'Fx(n) Pstart(16px)')
+    top10 = parse_rank_list(top_block, 'Fx(n) Pstart(16px)', limit=10)
+    industries = parse_industries(industry_block)
+
+    return {
+        "symbol": symbol,
+        "url": url,
+        "asset_date": extract_date(asset_block),
+        "assets": assets,
+        "top10_date": extract_date(top_block),
+        "top10": top10,
+        "industry_date": extract_date(industry_block),
+        "industries": industries,
+    }
+
+
+def print_section(title: str, date: str, items: list[dict[str, str]]) -> None:
+    print(f"{title}｜資料時間：{date or '無資料'}")
+    for item in items:
+        print(f"{item['排名']}. {item['名稱']}：{item['占比']}")
+    print()
+
+
+def write_json(data: dict, out: str | None) -> None:
+    text = json.dumps(data, ensure_ascii=False, indent=2)
+    if out:
+        with open(out, 'w', encoding='utf-8') as f:
+            f.write(text)
+    else:
+        print(text)
+
+
+def write_csv(data: dict, out: str | None, section: str) -> None:
+    mapping = {
+        'top': ('top10', 'top10_date'),
+        'industry': ('industries', 'industry_date'),
+        'asset': ('assets', 'asset_date'),
+    }
+    rows = []
+    if section == 'all':
+        for sec in ['top', 'industry', 'asset']:
+            key, date_key = mapping[sec]
+            rows.extend({"section": sec, "data_date": data[date_key], **item} for item in data[key])
+    else:
+        key, date_key = mapping[section]
+        rows.extend({"section": section, "data_date": data[date_key], **item} for item in data[key])
+
+    fieldnames = ['section', 'data_date', '排名', '名稱', '占比']
+    if out:
+        f = open(out, 'w', encoding='utf-8-sig', newline='')
+    else:
+        f = sys.stdout
+    try:
+        writer = csv.DictWriter(f, fieldnames=fieldnames)
+        writer.writeheader()
+        writer.writerows(rows)
+    finally:
+        if out:
+            f.close()
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(description='抓 Yahoo ETF 持股分析頁資料')
+    parser.add_argument('symbol', help='ETF 代號，例如 00922、0050')
+    parser.add_argument('--format', choices=['text', 'json', 'csv'], default='text')
+    parser.add_argument('--section', choices=['all', 'top', 'industry', 'asset'], default='all')
+    parser.add_argument('--out', help='輸出檔案路徑')
+    args = parser.parse_args()
+
+    data = parse_holding_page(args.symbol)
+
+    if args.format == 'json':
+        write_json(data, args.out)
+        return
+    if args.format == 'csv':
+        write_csv(data, args.out, args.section)
+        return
+
+    if args.section in ('all', 'top'):
+        print_section(f"{args.symbol} 前十大持股", data['top10_date'], data['top10'])
+    if args.section in ('all', 'industry'):
+        print_section(f"{args.symbol} 行業比重", data['industry_date'], data['industries'])
+    if args.section in ('all', 'asset'):
+        print_section(f"{args.symbol} 資產分佈", data['asset_date'], data['assets'])
+
+
+if __name__ == '__main__':
+    main()
--- a/scripts/yahoo-institutional-browserless.py
+++ b/scripts/yahoo-institutional-browserless.py
@@ -0,0 +1,158 @@
+#!/usr/bin/env python3
+"""
+用 Browserless 抓 Yahoo 三大法人頁，輸出精簡摘要（代號可變）
+
+用法：
+  python3 stocks/yahoo-institutional-browserless.py 3481
+  python3 stocks/yahoo-institutional-browserless.py 2330
+"""
+
+import json
+import os
+from pathlib import Path
+import re
+import sys
+import subprocess
+
+
+
+def load_dotenv() -> None:
+    """載入 .env（優先目前目錄，其次 workspace 根目錄）。"""
+    candidates = [
+        Path.cwd() / ".env",
+        Path(__file__).resolve().parents[3] / ".env",
+    ]
+    for env_path in candidates:
+        if not env_path.exists():
+            continue
+        for raw in env_path.read_text(encoding="utf-8").splitlines():
+            line = raw.strip()
+            if not line or line.startswith("#") or "=" not in line:
+                continue
+            k, v = line.split("=", 1)
+            k = k.strip()
+            v = v.strip().strip('"').strip("'")
+            if k and k not in os.environ:
+                os.environ[k] = v
+        break
+
+
+load_dotenv()
+ENDPOINT = os.getenv("BROWSERLESS_ENDPOINT")
+TOKEN = os.getenv("BROWSERLESS_TOKEN")
+if not ENDPOINT or not TOKEN:
+    raise RuntimeError("缺少必要環境變數：BROWSERLESS_ENDPOINT / BROWSERLESS_TOKEN")
+STOCK_NO = sys.argv[1] if len(sys.argv) > 1 else "3481"
+
+
+def fetch_html(stock_no: str) -> str:
+    url = f"https://tw.stock.yahoo.com/quote/{stock_no}.TW/institutional-trading"
+    api = f"{ENDPOINT}/content?token={TOKEN}"
+    payload = json.dumps(
+        {
+            "url": url,
+            "gotoOptions": {"waitUntil": "domcontentloaded", "timeout": 60000},
+        },
+        ensure_ascii=False,
+    )
+
+    cmd = [
+        "curl",
+        "-sS",
+        "--max-time",
+        "90",
+        "-X",
+        "POST",
+        api,
+        "-H",
+        "Content-Type: application/json",
+        "-d",
+        payload,
+    ]
+    try:
+        text = subprocess.check_output(cmd, text=True, timeout=95)
+    except subprocess.CalledProcessError as e:
+        raise RuntimeError(f"Browserless 呼叫失敗: {e}") from e
+    except subprocess.TimeoutExpired as e:
+        raise RuntimeError("Browserless 呼叫逾時") from e
+
+    if "TimeoutError" in text:
+        raise RuntimeError("Browserless 導航逾時（TimeoutError）")
+    return text
+
+
+def pick(html: str, pattern: str, default: str = "-") -> str:
+    m = re.search(pattern, html, flags=re.S)
+    return m.group(1) if m else default
+
+
+def parse_inst(html: str, stock_no: str) -> dict:
+    name = pick(html, r'property="og:title" content="([^"(]+)\([0-9]+\.TW\)', "").strip()
+    if not name:
+        name = pick(html, r"<title>([^<(]+)\([0-9]+\.TW\)", "-").strip()
+
+    # 抓 summary data（避免大範圍 regex 回溯，先縮小搜尋範圍）
+    anchor = html.find('institutionBuySellSummaryData')
+    chunk = html[anchor: anchor + 250000] if anchor != -1 else html
+
+    rows = []
+    obj_pat = re.compile(r'\{[^{}]*"formattedDate":"[^"]+"[^{}]*\}')
+    for om in obj_pat.finditer(chunk):
+        s = om.group(0)
+        foreign_m = re.search(r'"foreignDiffVolK":(-?\d+)', s)
+        trust_m = re.search(r'"investmentTrustDiffVolK":(-?\d+)', s)
+        dealer_m = re.search(r'"dealerDiffVolK":(-?\d+)', s)
+        total_m = re.search(r'"totalDiffVolK":(-?\d+)', s)
+        date_m = re.search(r'"formattedDate":"([^"]+)"', s)
+        if foreign_m and trust_m and dealer_m and total_m and date_m:
+            rows.append((foreign_m.group(1), trust_m.group(1), dealer_m.group(1), total_m.group(1), date_m.group(1)))
+
+    if not rows:
+        raise RuntimeError("抓不到三大法人資料（頁面結構可能變更）")
+
+    def normalize_date(ds: str) -> str:
+        return ds.replace("\\u002F", "/")
+
+    def date_key(ds: str):
+        s = normalize_date(ds)
+        try:
+            y, m, d = s.split("/")
+            return int(y), int(m), int(d)
+        except Exception:
+            return (0, 0, 0)
+
+    foreign, trust, dealer, total, date = max(rows, key=lambda r: date_key(r[4]))
+    date = normalize_date(date)
+
+    return {
+        "name": name,
+        "stock_no": stock_no,
+        "date": date,
+        "foreign": int(foreign),
+        "trust": int(trust),
+        "dealer": int(dealer),
+        "total": int(total),
+    }
+
+
+def sign(v: int) -> str:
+    return f"+{v:,}" if v >= 0 else f"{v:,}"
+
+
+def print_summary(d: dict) -> None:
+    print(f"{d['name']}（{d['stock_no']}）｜法人買賣超摘要")
+    print(f"• 日期：{d['date']}")
+    print(f"• 外資：{sign(d['foreign'])} 張")
+    print(f"• 投信：{sign(d['trust'])} 張")
+    print(f"• 自營商：{sign(d['dealer'])} 張")
+    print(f"• 三大法人合計：{sign(d['total'])} 張")
+
+
+if __name__ == "__main__":
+    try:
+        html = fetch_html(STOCK_NO)
+        data = parse_inst(html, STOCK_NO)
+        print_summary(data)
+    except Exception as e:
+        print(f"取得失敗：{e}", file=sys.stderr)
+        sys.exit(1)
--- a/scripts/yahoo-ma-signal.py
+++ b/scripts/yahoo-ma-signal.py
@@ -0,0 +1,82 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+
+from __future__ import annotations
+
+import json
+import sys
+import urllib.parse
+import urllib.request
+
+
+def fetch_closes(symbol: str, range_: str = "6mo", interval: str = "1d") -> list[float]:
+    base = f"https://query1.finance.yahoo.com/v8/finance/chart/{symbol}"
+    params = {"range": range_, "interval": interval}
+    url = f"{base}?{urllib.parse.urlencode(params)}"
+
+    req = urllib.request.Request(url, headers={"User-Agent": "Mozilla/5.0"})
+    with urllib.request.urlopen(req, timeout=30) as resp:
+        data = json.loads(resp.read().decode("utf-8", errors="replace"))
+
+    result = data.get("chart", {}).get("result", [])
+    if not result:
+        return []
+
+    quote = result[0].get("indicators", {}).get("quote", [])
+    if not quote:
+        return []
+
+    closes = quote[0].get("close", [])
+    return [float(x) for x in closes if x is not None]
+
+
+def sma(values: list[float], window: int) -> float:
+    return sum(values[-window:]) / window
+
+
+def main() -> None:
+    symbol = sys.argv[1] if len(sys.argv) > 1 else "00922.TW"
+    closes = fetch_closes(symbol)
+
+    if len(closes) < 60:
+        print(f"{symbol} 資料不足：至少需要 60 筆日收盤，當前 {len(closes)}")
+        return
+
+    price = closes[-1]
+    ma5 = sma(closes, 5)
+    ma10 = sma(closes, 10)
+    ma20 = sma(closes, 20)
+    ma60 = sma(closes, 60)
+
+    bias20 = ((price - ma20) / ma20) * 100
+
+    print(f"{symbol} 均線摘要")
+    print(f"- 收盤：{price:.2f}")
+    print(f"- MA5 ：{ma5:.2f}")
+    print(f"- MA10：{ma10:.2f}")
+    print(f"- MA20：{ma20:.2f}")
+    print(f"- MA60：{ma60:.2f}")
+    print(f"- 20日乖離：{bias20:.2f}%")
+
+    # 簡單判讀
+    trend = []
+    if price > ma20:
+        trend.append("站上月線")
+    else:
+        trend.append("跌破月線")
+
+    if ma20 > ma60:
+        trend.append("中期偏多")
+    else:
+        trend.append("中期偏弱")
+
+    if ma5 > ma20:
+        trend.append("短線強於月線")
+    else:
+        trend.append("短線弱於月線")
+
+    print(f"- 判讀：{'／'.join(trend)}")
+
+
+if __name__ == "__main__":
+    main()
--- a/scripts/yahoo-macd-signal.py
+++ b/scripts/yahoo-macd-signal.py
@@ -0,0 +1,91 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+
+from __future__ import annotations
+
+import json
+import sys
+import urllib.parse
+import urllib.request
+
+
+def fetch_closes(symbol: str, range_: str = "6mo", interval: str = "1d") -> list[float]:
+    base = f"https://query1.finance.yahoo.com/v8/finance/chart/{symbol}"
+    params = {"range": range_, "interval": interval}
+    url = f"{base}?{urllib.parse.urlencode(params)}"
+
+    req = urllib.request.Request(url, headers={"User-Agent": "Mozilla/5.0"})
+    with urllib.request.urlopen(req, timeout=30) as resp:
+        data = json.loads(resp.read().decode("utf-8", errors="replace"))
+
+    result = data.get("chart", {}).get("result", [])
+    if not result:
+        return []
+
+    quote = result[0].get("indicators", {}).get("quote", [])
+    if not quote:
+        return []
+
+    closes = quote[0].get("close", [])
+    return [float(x) for x in closes if x is not None]
+
+
+def ema(values: list[float], period: int) -> list[float]:
+    if not values:
+        return []
+    k = 2 / (period + 1)
+    out = [values[0]]
+    for v in values[1:]:
+        out.append(v * k + out[-1] * (1 - k))
+    return out
+
+
+def macd(closes: list[float]) -> tuple[list[float], list[float], list[float]]:
+    ema12 = ema(closes, 12)
+    ema26 = ema(closes, 26)
+    dif = [a - b for a, b in zip(ema12, ema26)]
+    dea = ema(dif, 9)
+    hist = [d - s for d, s in zip(dif, dea)]
+    return dif, dea, hist
+
+
+def main() -> None:
+    symbol = sys.argv[1] if len(sys.argv) > 1 else "00922.TW"
+    closes = fetch_closes(symbol)
+
+    if len(closes) < 35:
+        print(f"{symbol} 資料不足，無法判斷 MACD（至少 35 筆，當前 {len(closes)}）")
+        return
+
+    dif, dea, hist = macd(closes)
+
+    d0, d1 = dif[-1], dif[-2]
+    s0, s1 = dea[-1], dea[-2]
+    h0, h1 = hist[-1], hist[-2]
+
+    golden_cross = d1 <= s1 and d0 > s0
+    death_cross = d1 >= s1 and d0 < s0
+    low_zone = d0 < 0 and s0 < 0
+
+    if golden_cross and low_zone:
+        signal = "✅ 低位金叉：偏多訊號"
+    elif golden_cross:
+        signal = "✅ 金叉：短線轉強"
+    elif death_cross:
+        signal = "⚠️ 死叉：短線轉弱"
+    elif h0 > h1 and h0 > 0:
+        signal = "📈 多頭延續（柱體擴大）"
+    elif h0 < h1 and h0 < 0:
+        signal = "📉 空頭延續（柱體擴大）"
+    else:
+        signal = "👀 MACD 震盪中，先觀察"
+
+    print(f"{symbol} MACD 摘要")
+    print(f"- DIF：{d0:.4f}")
+    print(f"- DEA：{s0:.4f}")
+    print(f"- HIST：{h0:.4f}")
+    print(f"- 訊號：{signal}")
+
+
+if __name__ == "__main__":
+    main()
--- a/scripts/yahoo-market-brief.py
+++ b/scripts/yahoo-market-brief.py
@@ -0,0 +1 @@
+yahoo-market-browserless.py
--- a/scripts/yahoo-market-browserless.py
+++ b/scripts/yahoo-market-browserless.py
@@ -0,0 +1,168 @@
+#!/usr/bin/env python3
+"""
+用 Browserless 抓 Yahoo 台股大盤（加權指數）即時行情。
+
+用法：
+  python3 stocks/yahoo-market-browserless.py
+  python3 stocks/yahoo-market-browserless.py ^TWII
+
+環境變數（可覆蓋）：
+  BROWSERLESS_ENDPOINT=http://192.168.42.124:13000
+  BROWSERLESS_TOKEN=6R0W53R135510
+"""
+
+import os
+import re
+import sys
+import json
+import urllib.request
+import urllib.error
+from pathlib import Path
+
+
+def load_dotenv() -> None:
+    """載入 .env（優先目前目錄，其次 workspace 根目錄）。"""
+    candidates = [
+        Path.cwd() / ".env",
+        Path(__file__).resolve().parents[3] / ".env",
+    ]
+    for env_path in candidates:
+        if not env_path.exists():
+            continue
+        for raw in env_path.read_text(encoding="utf-8").splitlines():
+            line = raw.strip()
+            if not line or line.startswith("#") or "=" not in line:
+                continue
+            k, v = line.split("=", 1)
+            k = k.strip()
+            v = v.strip().strip('"').strip("'")
+            if k and k not in os.environ:
+                os.environ[k] = v
+        break
+
+
+load_dotenv()
+
+ENDPOINT = os.getenv("BROWSERLESS_ENDPOINT")
+TOKEN = os.getenv("BROWSERLESS_TOKEN")
+if not ENDPOINT or not TOKEN:
+    raise RuntimeError("缺少必要環境變數：BROWSERLESS_ENDPOINT / BROWSERLESS_TOKEN")
+SYMBOL = sys.argv[1] if len(sys.argv) > 1 else "^TWII"
+
+
+def fetch_html(symbol: str) -> str:
+    url = f"https://tw.stock.yahoo.com/quote/{symbol}"
+    api = f"{ENDPOINT}/content?token={TOKEN}"
+    payload = {
+        "url": url,
+        "gotoOptions": {"waitUntil": "domcontentloaded", "timeout": 60000},
+    }
+
+    req = urllib.request.Request(
+        api,
+        data=json.dumps(payload).encode("utf-8"),
+        headers={"Content-Type": "application/json"},
+        method="POST",
+    )
+
+    try:
+        with urllib.request.urlopen(req, timeout=90) as resp:
+            text = resp.read().decode("utf-8", errors="replace")
+            status = resp.getcode()
+    except urllib.error.HTTPError as e:
+        body = e.read().decode("utf-8", errors="replace") if hasattr(e, "read") else str(e)
+        raise RuntimeError(f"Browserless HTTP {e.code}: {body[:200]}") from e
+
+    if status != 200:
+        raise RuntimeError(f"Browserless HTTP {status}: {text[:200]}")
+    if "TimeoutError" in text:
+        raise RuntimeError("Browserless 導航逾時（TimeoutError）")
+    return text
+
+
+def pick(html: str, pattern: str, default: str = "-") -> str:
+    m = re.search(pattern, html, flags=re.S)
+    return m.group(1) if m else default
+
+
+def to_float(num: str) -> float | None:
+    try:
+        if not num or num == "-":
+            return None
+        return float(num.replace(",", ""))
+    except Exception:
+        return None
+
+
+def parse_market(html: str, symbol: str) -> dict:
+    # 名稱：優先取 title（加權指數(^TWII) 走勢圖 - Yahoo股市）
+    name = pick(html, r"<title>([^<(]+)\([^)]*\)\s*走勢圖")
+    if name == "-":
+        name = pick(html, r'property="og:title" content="([^<(]+)\([^)]*\)')
+
+    price = pick(html, r"Fz\(32px\)[^>]*>([\d,]+\.?\d*)<")
+
+    # 漲跌值常出現在百分比前一個 span，且前面有紅綠三角形 span
+    updown = pick(
+        html,
+        r'border-width:[^"]+\"></span>([+\-]?[\d,]+\.?\d*)</span><span[^>]*>\([+\-]?[0-9]+\.[0-9]+%\)',
+    )
+    pct = pick(html, r"\(([+\-]?[0-9]+\.[0-9]+%)\)</span>")
+
+    open_price = pick(html, r">開盤</span><span[^>]*>([\d,]+\.?\d*)")
+    high = pick(html, r">最高</span><span[^>]*>([\d,]+\.?\d*)")
+    low = pick(html, r">最低</span><span[^>]*>([\d,]+\.?\d*)")
+    prev = pick(html, r">昨收</span><span[^>]*>([\d,]+\.?\d*)")
+
+    amount_100m = pick(html, r">成交金額\(億\)</span><span[^>]*>([\d,]+\.?\d*)")
+    time_raw = pick(html, r"(?:開盤|收盤) \| ([0-9/ :]+) 更新")
+    time_fmt = "-" if time_raw == "-" else f"{time_raw.replace('/', '-')}（台北）"
+
+    # 以指數與昨收重算漲跌，避免頁面符號誤判
+    p = to_float(price)
+    pv = to_float(prev)
+    if p is not None and pv is not None:
+        diff = p - pv
+        updown = f"{diff:+.2f}"
+        if pv != 0:
+            pct = f"{(diff / pv) * 100:+.2f}%"
+
+    return {
+        "name": name,
+        "symbol": symbol,
+        "price": price,
+        "updown": updown,
+        "pct": pct,
+        "open": open_price,
+        "high": high,
+        "low": low,
+        "prev": prev,
+        "amount_100m": amount_100m,
+        "time": time_fmt,
+    }
+
+
+def sign_num(v: str) -> str:
+    if v in ("-", ""):
+        return "-"
+    return v if v.startswith("-") or v.startswith("+") else f"+{v}"
+
+
+def print_summary(q: dict) -> None:
+    print(f"{q['name']}（{q['symbol']}）")
+    print(f"• 指數：{q['price']}")
+    print(f"• 漲跌：{sign_num(q['updown'])}（{sign_num(q['pct'])}）")
+    print(f"• 開盤／最高／最低：{q['open']}／{q['high']}／{q['low']}")
+    print(f"• 昨收：{q['prev']}")
+    print(f"• 成交金額：{q['amount_100m']} 億")
+    print(f"• 資料時間：{q['time']}")
+
+
+if __name__ == "__main__":
+    try:
+        html = fetch_html(SYMBOL)
+        market = parse_market(html, SYMBOL)
+        print_summary(market)
+    except Exception as e:
+        print(f"取得失敗：{e}", file=sys.stderr)
+        sys.exit(1)
--- a/scripts/yahoo-quote-browserless.py
+++ b/scripts/yahoo-quote-browserless.py
@@ -0,0 +1,192 @@
+#!/usr/bin/env python3
+"""
+用 Browserless 抓 Yahoo 股市頁面，輸出精簡報價摘要（省 token 版）
+
+用法：
+  python3 stocks/yahoo-quote-browserless.py 3481
+  python3 stocks/yahoo-quote-browserless.py 2330
+
+環境變數（可覆蓋）：
+  BROWSERLESS_ENDPOINT=http://192.168.42.124:13000
+  BROWSERLESS_TOKEN=6R0W53R135510
+"""
+
+import os
+import re
+import sys
+import json
+import urllib.request
+import urllib.error
+from pathlib import Path
+
+
+def load_dotenv() -> None:
+    """載入 .env（優先目前目錄，其次 workspace 根目錄）。"""
+    candidates = [
+        Path.cwd() / ".env",
+        Path(__file__).resolve().parents[3] / ".env",
+    ]
+    for env_path in candidates:
+        if not env_path.exists():
+            continue
+        for raw in env_path.read_text(encoding="utf-8").splitlines():
+            line = raw.strip()
+            if not line or line.startswith("#") or "=" not in line:
+                continue
+            k, v = line.split("=", 1)
+            k = k.strip()
+            v = v.strip().strip('"').strip("'")
+            if k and k not in os.environ:
+                os.environ[k] = v
+        break
+
+
+load_dotenv()
+
+ENDPOINT = os.getenv("BROWSERLESS_ENDPOINT")
+TOKEN = os.getenv("BROWSERLESS_TOKEN")
+if not ENDPOINT or not TOKEN:
+    raise RuntimeError("缺少必要環境變數：BROWSERLESS_ENDPOINT / BROWSERLESS_TOKEN")
+STOCK_NO = sys.argv[1] if len(sys.argv) > 1 else "3481"
+
+
+def fetch_html(stock_no: str) -> str:
+    url = f"https://tw.stock.yahoo.com/quote/{stock_no}.TW"
+    api = f"{ENDPOINT}/content?token={TOKEN}"
+    payload = {
+        "url": url,
+        "gotoOptions": {"waitUntil": "domcontentloaded", "timeout": 60000},
+    }
+
+    req = urllib.request.Request(
+        api,
+        data=json.dumps(payload).encode("utf-8"),
+        headers={"Content-Type": "application/json"},
+        method="POST",
+    )
+
+    try:
+        with urllib.request.urlopen(req, timeout=90) as resp:
+            text = resp.read().decode("utf-8", errors="replace")
+            status = resp.getcode()
+    except urllib.error.HTTPError as e:
+        body = e.read().decode("utf-8", errors="replace") if hasattr(e, "read") else str(e)
+        raise RuntimeError(f"Browserless HTTP {e.code}: {body[:200]}") from e
+
+    if status != 200:
+        raise RuntimeError(f"Browserless HTTP {status}: {text[:200]}")
+    if "TimeoutError" in text:
+        raise RuntimeError("Browserless 導航逾時（TimeoutError）")
+    return text
+
+
+def pick(html: str, pattern: str, default: str = "-") -> str:
+    m = re.search(pattern, html, flags=re.S)
+    return m.group(1) if m else default
+
+
+def to_float(num: str) -> float | None:
+    try:
+        if not num or num == "-":
+            return None
+        return float(num.replace(",", ""))
+    except Exception:
+        return None
+
+
+def parse_quote(html: str, stock_no: str) -> dict:
+    # 名稱容錯：有些股票 h1 結構會不一樣
+    name = pick(html, r"<h1[^>]*>([^<(]+)", "-").strip()
+    if not name or name in ("-", "Yahoo股市"):
+        name = pick(html, r'property="og:title" content="([^"(]+)\([0-9]+\.TW\)', "").strip()
+    if not name or name in ("-", "Yahoo股市"):
+        name = pick(html, r"<title>([^<(]+)\([0-9]+\.TW\)", "-").strip()
+
+    # 價格容錯：支援千分位與不同 class 變化
+    num_pat = r"([\d,]+\.?\d*)"
+    price = pick(html, rf"Fz\(32px\)[^>]*>{num_pat}<")
+
+    # 漲跌先抓頁面值，後續再用（現價-昨收）覆寫，避免符號抓錯
+    updown = pick(html, rf"Fz\(20px\)[^>]*>([\d\.\+\-]+)<", "-")
+    if updown == "-":
+        updown = pick(html, rf"Fz\(20px\)[^>]*>[\s\S]*?</span>([\d\.\+\-]+)<", "0.00")
+
+    pct = pick(html, r"\(([+-]?[0-9]+\.[0-9]+%)\)</span>", "0.00%")
+
+    open_price = pick(html, rf">開盤</span><span[^>]*>{num_pat}")
+    high = pick(html, rf">最高</span><span[^>]*>{num_pat}")
+    low = pick(html, rf">最低</span><span[^>]*>{num_pat}")
+    prev = pick(html, rf">昨收</span><span[^>]*>{num_pat}")
+    avg = pick(html, rf">均價</span><span[^>]*>{num_pat}")
+
+    volume_lots = pick(
+        html,
+        r'Mb\(4px\)">([0-9,]+)</span><span class="Fz\(12px\) C\(\$c-icon\)">成交量',
+        "0",
+    )
+    volume_shares = "{:,.0f}".format(float(volume_lots.replace(",", "")) * 1000)
+
+    bids = re.findall(r"Fw\(n\)[^>]*>([\d,]+\.?\d*)</span>", html)
+    best_bid = bids[0] if len(bids) >= 1 else "-"
+    best_ask = bids[5] if len(bids) >= 6 else "-"
+
+    time_raw = pick(html, r"開盤 \| ([0-9/ :]+) 更新")
+    time_fmt = "-" if time_raw == "-" else f"{time_raw.replace('/', '-')}（台北）"
+    market = "未知" if time_raw == "-" else "開盤中"
+
+    # 以現價與昨收重算漲跌，避免頁面箭頭造成正負號誤判
+    p = to_float(price)
+    pv = to_float(prev)
+    if p is not None and pv is not None:
+        diff = p - pv
+        updown = f"{diff:+.2f}"
+        if pv != 0:
+            pct = f"{(diff / pv) * 100:+.2f}%"
+
+    return {
+        "name": name,
+        "stock_no": stock_no,
+        "price": price,
+        "updown": updown,
+        "pct": pct,
+        "open": open_price,
+        "high": high,
+        "low": low,
+        "prev": prev,
+        "volume_lots": volume_lots,
+        "volume_shares": volume_shares,
+        "avg": avg,
+        "best_bid": best_bid,
+        "best_ask": best_ask,
+        "market": market,
+        "time": time_fmt,
+    }
+
+
+def sign_num(v: str) -> str:
+    if v in ("-", ""):
+        return "-"
+    return v if v.startswith("-") or v.startswith("+") else f"+{v}"
+
+
+def print_summary(q: dict) -> None:
+    print(f"{q['name']}（{q['stock_no']}）")
+    print(f"• 現價：{q['price']}")
+    print(f"• 漲跌：{sign_num(q['updown'])}（{sign_num(q['pct'])}）")
+    print(f"• 開盤／最高／最低：{q['open']}／{q['high']}／{q['low']}")
+    print(f"• 昨收：{q['prev']}")
+    print(f"• 成交量：{q['volume_shares']}（Yahoo 顯示 {q['volume_lots']} 張，換算股數）")
+    print(f"• 均價：{q['avg']}")
+    print(f"• 最佳買賣：{q['best_bid']}／{q['best_ask']}")
+    print(f"• 市場狀態：{q['market']}")
+    print(f"• 資料時間：{q['time']}")
+
+
+if __name__ == "__main__":
+    try:
+        html = fetch_html(STOCK_NO)
+        quote = parse_quote(html, STOCK_NO)
+        print_summary(quote)
+    except Exception as e:
+        print(f"取得失敗：{e}", file=sys.stderr)
+        sys.exit(1)
--- a/scripts/yahoo-technical-browserless.py
+++ b/scripts/yahoo-technical-browserless.py
@@ -0,0 +1,303 @@
+#!/usr/bin/env python3
+"""
+用 Browserless 透過 Yahoo Chart API 抓技術面摘要（代號可變）
+
+輸出內容：
+- 近 5 個交易日 K 線（開高低收）
+- RSI14
+- ATR14
+- KD（9,3,3）
+- 布林通道（20,2）
+
+用法：
+  python3 skills/stocks-query/scripts/yahoo-technical-browserless.py 3481
+  python3 skills/stocks-query/scripts/yahoo-technical-browserless.py 2330
+"""
+
+import datetime as dt
+import html as html_lib
+import json
+import math
+import os
+from pathlib import Path
+import re
+import subprocess
+import sys
+
+
+def load_dotenv() -> None:
+    """載入 .env（優先目前目錄，其次 workspace 根目錄）。"""
+    candidates = [
+        Path.cwd() / ".env",
+        Path(__file__).resolve().parents[3] / ".env",
+    ]
+    for env_path in candidates:
+        if not env_path.exists():
+            continue
+        for raw in env_path.read_text(encoding="utf-8").splitlines():
+            line = raw.strip()
+            if not line or line.startswith("#") or "=" not in line:
+                continue
+            k, v = line.split("=", 1)
+            k = k.strip()
+            v = v.strip().strip('"').strip("'")
+            if k and k not in os.environ:
+                os.environ[k] = v
+        break
+
+
+load_dotenv()
+ENDPOINT = os.getenv("BROWSERLESS_ENDPOINT")
+TOKEN = os.getenv("BROWSERLESS_TOKEN")
+if not ENDPOINT or not TOKEN:
+    raise RuntimeError("缺少必要環境變數：BROWSERLESS_ENDPOINT / BROWSERLESS_TOKEN")
+STOCK_NO = sys.argv[1] if len(sys.argv) > 1 else "3481"
+
+
+def fetch_chart_json(stock_no: str) -> dict:
+    api_url = f"https://query1.finance.yahoo.com/v8/finance/chart/{stock_no}.TW?range=3mo&interval=1d"
+    browserless_api = f"{ENDPOINT}/content?token={TOKEN}"
+    payload = json.dumps(
+        {
+            "url": api_url,
+            "gotoOptions": {"waitUntil": "domcontentloaded", "timeout": 60000},
+        },
+        ensure_ascii=False,
+    )
+
+    cmd = [
+        "curl",
+        "-sS",
+        "--max-time",
+        "90",
+        "-X",
+        "POST",
+        browserless_api,
+        "-H",
+        "Content-Type: application/json",
+        "-d",
+        payload,
+    ]
+
+    try:
+        raw = subprocess.check_output(cmd, text=True)
+    except subprocess.CalledProcessError as e:
+        raise RuntimeError(f"Browserless 呼叫失敗: {e}") from e
+    except subprocess.TimeoutExpired as e:
+        raise RuntimeError("Browserless 呼叫逾時") from e
+
+    m = re.search(r"<pre>([\s\S]+)</pre>", raw)
+    json_text = html_lib.unescape(m.group(1) if m else raw).strip()
+
+    data = json.loads(json_text)
+    if data.get("chart", {}).get("error"):
+        raise RuntimeError(f"Yahoo API 錯誤: {data['chart']['error']}")
+    return data
+
+
+def compute_rsi14(closes: list[float]) -> float | None:
+    if len(closes) < 15:
+        return None
+
+    gains = []
+    losses = []
+    for prev_close, close in zip(closes[:-1], closes[1:]):
+        change = close - prev_close
+        gains.append(max(change, 0.0))
+        losses.append(max(-change, 0.0))
+
+    period = 14
+    avg_gain = sum(gains[:period]) / period
+    avg_loss = sum(losses[:period]) / period
+
+    for gain, loss in zip(gains[period:], losses[period:]):
+        avg_gain = ((avg_gain * (period - 1)) + gain) / period
+        avg_loss = ((avg_loss * (period - 1)) + loss) / period
+
+    if avg_loss == 0:
+        return 100.0
+
+    rs = avg_gain / avg_loss
+    return 100 - (100 / (1 + rs))
+
+
+def compute_atr14(highs: list[float], lows: list[float], closes: list[float]) -> float | None:
+    if len(highs) < 15 or len(lows) < 15 or len(closes) < 15:
+        return None
+
+    true_ranges = []
+    prev_close = closes[0]
+    for high, low, close in zip(highs[1:], lows[1:], closes[1:]):
+        tr = max(
+            high - low,
+            abs(high - prev_close),
+            abs(low - prev_close),
+        )
+        true_ranges.append(tr)
+        prev_close = close
+
+    period = 14
+    if len(true_ranges) < period:
+        return None
+
+    atr = sum(true_ranges[:period]) / period
+    for tr in true_ranges[period:]:
+        atr = ((atr * (period - 1)) + tr) / period
+    return atr
+
+
+def compute_kd(highs: list[float], lows: list[float], closes: list[float]) -> tuple[float | None, float | None]:
+    period = 9
+    if len(highs) < period or len(lows) < period or len(closes) < period:
+        return None, None
+
+    k_value = 50.0
+    d_value = 50.0
+    rsv_values = []
+
+    for idx in range(period - 1, len(closes)):
+        window_high = max(highs[idx - period + 1 : idx + 1])
+        window_low = min(lows[idx - period + 1 : idx + 1])
+        close = closes[idx]
+
+        if window_high == window_low:
+            rsv = 50.0
+        else:
+            rsv = ((close - window_low) / (window_high - window_low)) * 100
+        rsv_values.append(rsv)
+
+    for rsv in rsv_values:
+        k_value = ((2 / 3) * k_value) + ((1 / 3) * rsv)
+        d_value = ((2 / 3) * d_value) + ((1 / 3) * k_value)
+
+    return k_value, d_value
+
+
+def compute_bollinger_bands(closes: list[float]) -> tuple[float | None, float | None, float | None]:
+    period = 20
+    if len(closes) < period:
+        return None, None, None
+
+    window = closes[-period:]
+    middle = sum(window) / period
+    variance = sum((price - middle) ** 2 for price in window) / period
+    std_dev = math.sqrt(variance)
+    upper = middle + (2 * std_dev)
+    lower = middle - (2 * std_dev)
+    return upper, middle, lower
+
+
+def parse_technical_data(data: dict):
+    result = data["chart"]["result"][0]
+    symbol = result.get("meta", {}).get("symbol", f"{STOCK_NO}.TW")
+    timestamps = result.get("timestamp", [])
+    quote = result.get("indicators", {}).get("quote", [{}])[0]
+    opens = quote.get("open", [])
+    closes = quote.get("close", [])
+    highs = quote.get("high", [])
+    lows = quote.get("low", [])
+
+    kline_rows = []
+    close_values = []
+    high_values = []
+    low_values = []
+    for ts, open_price, close, high, low in zip(timestamps, opens, closes, highs, lows):
+        if None in (open_price, close, high, low):
+            continue
+        d = dt.datetime.fromtimestamp(ts, tz=dt.timezone(dt.timedelta(hours=8)))
+        open_float = float(open_price)
+        close_float = float(close)
+        high_float = float(high)
+        low_float = float(low)
+        kline_rows.append(
+            (d.strftime("%m/%d"), open_float, high_float, low_float, close_float)
+        )
+        close_values.append(close_float)
+        high_values.append(high_float)
+        low_values.append(low_float)
+
+    if len(kline_rows) < 5:
+        raise RuntimeError("可用 K 線資料不足 5 筆")
+
+    rsi14 = compute_rsi14(close_values)
+    atr14 = compute_atr14(high_values, low_values, close_values)
+    k_value, d_value = compute_kd(high_values, low_values, close_values)
+    boll_upper, boll_middle, boll_lower = compute_bollinger_bands(close_values)
+    return (
+        symbol,
+        kline_rows[-5:],
+        rsi14,
+        atr14,
+        k_value,
+        d_value,
+        boll_upper,
+        boll_middle,
+        boll_lower,
+    )
+
+
+def print_summary(
+    stock_no: str,
+    symbol: str,
+    kline_rows,
+    rsi14: float | None,
+    atr14: float | None,
+    k_value: float | None,
+    d_value: float | None,
+    boll_upper: float | None,
+    boll_middle: float | None,
+    boll_lower: float | None,
+):
+    print(f"{stock_no}（{symbol}）｜技術面摘要")
+    print("• 近 5 個交易日 K 線（開／高／低／收）")
+    for d, o, h, l, c in kline_rows:
+        print(f"  - {d}：{o:.2f}／{h:.2f}／{l:.2f}／{c:.2f}")
+    if rsi14 is None:
+        print("• RSI14：無資料")
+    else:
+        print(f"• RSI14：{rsi14:.2f}")
+    if atr14 is None:
+        print("• ATR14：無資料")
+    else:
+        print(f"• ATR14：{atr14:.2f}")
+    if k_value is None or d_value is None:
+        print("• KD：無資料")
+    else:
+        print(f"• K：{k_value:.2f}")
+        print(f"• D：{d_value:.2f}")
+    if boll_upper is None or boll_middle is None or boll_lower is None:
+        print("• 布林通道：無資料")
+    else:
+        print(f"• 布林上軌：{boll_upper:.2f}")
+        print(f"• 布林中軌：{boll_middle:.2f}")
+        print(f"• 布林下軌：{boll_lower:.2f}")
+
+
+if __name__ == "__main__":
+    try:
+        (
+            symbol,
+            kline_rows,
+            rsi14,
+            atr14,
+            k_value,
+            d_value,
+            boll_upper,
+            boll_middle,
+            boll_lower,
+        ) = parse_technical_data(fetch_chart_json(STOCK_NO))
+        print_summary(
+            STOCK_NO,
+            symbol,
+            kline_rows,
+            rsi14,
+            atr14,
+            k_value,
+            d_value,
+            boll_upper,
+            boll_middle,
+            boll_lower,
+        )
+    except Exception as e:
+        print(f"取得失敗：{e}", file=sys.stderr)
+        sys.exit(1)