SuperAPI AI Directory
Partner中文AI newsFavorites
Skillsstock-feed

stock-feed

General & other

View repository

Updated 10-16 01:48

About this skill

A-share daily news

📝 Overview

A-share daily news is a Chinese stock market public opinion research tool. It searches 17 core A-share keywords at once from the three major platforms Xiaohongshu, Douyin, and WeChat Official Accounts (A-share, A-share market, A-share broad market, limit-up, rise/fall, potential stocks, stock selection, adding positions, etc.), pulls real discussion data from the past 7 days by default, automatically filters out non-stock/investment-related content, performs cross-platform comparative analysis of stock market public opinion trends, and outputs structured research reports and visualized HTML.

✨ Features

Function Module Capability Description Core Value
17-word one-click query Built-in core A-share keywords, covering all topics at once Zero-configuration ready to use
Three-platform data sources Real data from Xiaohongshu, Douyin, and WeChat Official Accounts Multi-dimensional public opinion perspectives
Cross-platform comparison Automatically synthesizes data from the three platforms to generate insights Discover differentiated trends
Flexible time range Defaults to 7 days, supports custom 1–30 days Track intraday/weekly changes
HTML report Interactive visualized report generation Easy to share and spread
Historical backtracking Supports data for any date within the past 30 days Track trend evolution

🔑 Authentication

The script requires an API Key to be configured. Obtain a personal Key from Redfox Data and configure the environment variable:

export REDFOX_API_KEY=ak_你的密钥

Priority: command line --api-key > REDFOX_API_KEY / X_API_KEY environment variable > configuration file.

Core Parameters

Parameter Description Default Value
keyword Search keywords (optional; if not provided, the built-in 17 A-share keywords are used) A-share, A-share market, A-share broad market, ... 17 words in total
--platforms Platform list (reducing is not recommended) xhs,dy,gzh
--count Number of items per platform 50
--days Time range (defaults to 7 days; pass as needed when specified by the user) 7
--output-format json / html / both json
--output-dir Output directory ~/Downloads/StockFeed

Platform signal interpretation:

Platform Content Characteristics Key Metrics
Xiaohongshu (xhs) Retail investor sharing, stock trading insights, beginner tutorials High save/like ratio = practical signal
Douyin (dy) Intraday quick commentary, limit-up interpretation, sentiment spread Number of shares =传播力
WeChat Official Accounts (gzh) In-depth review, strategy analysis, industry research reports Read count = attention, shares = recognition

Workflow

1. Environment check

Before running, ensure that the API Key has been configured (environment variable REDFOX_API_KEY or --api-key parameter). If not configured, prompt the user:

Please configure API Key:export REDFOX_API_KEY=ak_你的密钥, registration address https://www.redfox.hk/login

2. Keyword processing

  • Default behavior: If the keyword parameter is not passed, the script automatically uses the 17 built-in core A-share keywords
  • User customization: When the user explicitly specifies keywords, pass in the --keyword parameter to override the default value
  • Check whether the keyword is too vague; if so, ask the user to make it more specific

3. Pre-research

Before calling the engine, run 2-3 WebSearch queries in parallel to extract hot words and background:

  • A股 {TOPIC} 小红书 热门讨论 → extract Xiaohongshu hot words
  • A股 {TOPIC} 抖音 热门话题 → extract Douyin tags
  • A股 {TOPIC} 最新动态 → supplement timely background

4. Run the engine

python3 scripts/stock_feed.py \
  --platforms xhs,dy,gzh \
  --count 50 \
  --days 7 \
  --output-format json \
  --output-dir="${STOCK_FEED_MEMORY_DIR:-$HOME/Downloads/StockFeed}"

When the user specifies keywords:

python3 scripts/stock_feed.py "半导体,芯片" \
  --platforms xhs,dy,gzh \
  --count 50 \
  --days 7 \
  --output-format json \
  --output-dir="${STOCK_FEED_MEMORY_DIR:-$HOME/Downloads/StockFeed}"

Run in the foreground with a 5-minute timeout; do not run in the background. Read the full output, including titles, authors, engagement data, and links from the three platforms.

5. WebSearch supplement

After the engine is complete, supplement with 1-2 WebSearch queries covering sources not covered by the engine, such as Xueqiu, Eastmoney, and Tonghuashun. Exclude xiaohongshu.com / douyin.com / mp.weixin.qq.com (already covered by the engine).

6. Comprehensive output

Follow the output rules to generate the research report. Core principles:

  • After the badge and before the title, the "Data Overview" module must be output: in the order of WeChat Official Account → Xiaohongshu → Douyin, each platform displays a table of the TOP5 works (title as a clickable hyperlink, author, and platform-specific engagement metrics displayed in separate columns (WeChat Official Account reads/likes/shares, Xiaohongshu likes/favorites/comments, Douyin likes/comments/shares)), and at the end of the module display the total data volume and the HTML report prompt
  • The report starts with the first-level heading # A股每日新闻 begins
  • Synthesize by topic/story, not by listing by platform
  • Conclusions cross-validated across multiple platforms have the highest confidence
  • High-engagement content has the highest weight (including real user signals)
  • Each narrative paragraph starts with a bold heading, followed by - and the main text
  • Each finding must cite sources from at least 2 different platforms(e.g., Xiaohongshu + Douyin, WeChat Official Account + Xiaohongshu)
  • Citations must be clickable Markdown links
  • After the core findings, two fixed sections must be attached: "Operational Key Points Forecast" and "Risk Statement"

After the comprehensive report is completed,automatically executethe following steps to generate the HTML report (no need to ask the user):

# 直接从 JSON 生成 HTML（不调 API，不写入 Markdown）
python3 scripts/stock_feed.py --from-json "JSON文件路径" \
  --output-dir="${STOCK_FEED_MEMORY_DIR:-$HOME/Downloads/StockFeed}"

# 自动打开 HTML 报告
open "HTML文件路径"

The script will output the HTML file path; after automatically opening it, simply inform the user of the path.

Other resources