Daily News

Daily News

Overview

This skill provides tools to fetch daily news, weather, history events, idioms, poems, and daily quotes from the topurl.cn API and generate a plain text briefing. The briefing includes:

• Weather forecast for major cities
• Top 12 news stories intelligently selected and de-duplicated
• 3 historical events from this day
• Daily idiom (成语) with explanation
• Daily inspirational quote
• Daily progress bar for the current year

How It Works

Fetch News Data

Use fetch_news.py to retrieve full dataset from the API (only need to call once):

# Default: fetch full dataset with ip=202.106.0.20 (magic key)
python3 fetch_news.py

# Fetch with custom count (if needed)
python3 fetch_news.py --count 80

Important:

• Using --ip 202.106.0.20 is the "magic key" to get complete news and history data
• Default settings already include this IP, so no extra parameters needed
• Category filtering is done locally by generate_briefing.py, not by API
• Only call the API once to minimize server load

Generate Plain Text Briefing

Use generate_briefing.py to convert the JSON data into formatted plain text:

# First fetch the data
python3 fetch_news.py > news_data.json

# Then generate plain text from the JSON
python3 generate_briefing.py news_data.json

Important: The output is plain text (NOT HTML or Markdown), designed for easy copying and sharing in chat apps, emails, or social media.

Combined Workflow

For a complete workflow (recommended):

# One-time API call to fetch full dataset
python3 fetch_news.py > today.json

# Generate briefing from cached data (can run multiple times)
python3 generate_briefing.py today.json

You can also pipe the output:

python3 fetch_news.py | python3 generate_briefing.py

API Details

Endpoint

https://news.topurl.cn/api

Parameters

• ip: IP address for weather location (default: 202.106.0.20 - the "magic key" for full dataset)
• count: Number of news items to fetch (default: 60 for full dataset)

Note: The API is designed to be called once per day with the magic IP to get complete data. All filtering and processing should be done locally.

Response Structure

The API returns a JSON object with:

• calendar: Chinese lunar/civil calendar data
• newsList: Array of news items with title, URL, category, and importance score (0-100)
• historyList: Historical events from this day
• phrase: Daily idiom (成语) with explanation
• sentence: Daily inspirational quote
• weather: Weather information by city

News Selection Algorithm

The generate_briefing.py script uses an intelligent editorial algorithm:

• De-duplication: Removes similar news (60% title similarity threshold)
• Balanced sampling: Ensures diversity across categories (时事/国内/国际/商业)
• Score-based ranking: Prioritizes high-importance news (score 0-100)
• Corner case handling: Adapts when total news count is low (weekends)

Output Format

The script generates plain text in the following format:

慧语简报，1月29日星期四，农历腊月十一，工作愉快，平安喜乐

今天北京多云，-4 ~ 1℃，西南

1．特朗普酝酿对伊朗高层动手，伊朗警告发动史无前例报复；
2．缅北明家犯罪集团案11名罪犯被执行死刑；
3．中期选举亮红灯，特朗普就移民执法态度日趋软化；
...（共12条）

【历史上的今天】
1906年1月29日，电影在京城大受欢迎。
...（共3条）

【天天成语】
博而不精
释义：形容学识丰富，但不精深。
...

【慧语香风】
我们不但要提出任务，而且要解决完成任务的方法问题。
—— 毛泽东

【进度条】
▓░░░░░░░░░░░░░ 2026年，您已经使用了7.95%

(由 http://news.topurl.cn 采集整理)

DO NOT format the output as HTML or Markdown. The output must be plain text only. DO NOT add extra introductions or conclusions, DO NOT add unnecessary dividers or markdown formatting. Just output the briefing content directly.

Examples

Example 1: Standard Daily Workflow (Recommended)

# Step 1: Fetch full dataset (once per day)
python3 fetch_news.py > today.json

# Step 2: Generate briefing (can run multiple times)
python3 generate_briefing.py today.json

Example 2: Direct Output

# Fetch and generate in one command
python3 fetch_news.py | python3 generate_briefing.py

Resources

scripts/

• fetch_news.py - Fetch full dataset from the API (call once per day with magic IP)
• generate_briefing.py - Generate plain text briefing from JSON data (local processing)

references/

None

assets/

None

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: icheer
• Source: icheer/skills
• License: MIT

Install and usage instructions live in the source repository linked above.