Daily News Briefing Skill#

Automated daily news briefing system that generates comprehensive tech and financial reports with AI-powered commentary and delivers them via configured messaging channels. Supports dual-search architecture with Baidu API (preferred when configured) and DuckDuckGo fallback (always available, no API key required).

Installation & Configuration#

Step 1: Install via ClawHub#

Option A: From ClawHub (Published)

bash

# Search and install
clawhub search daily-news-briefing
clawhub install daily-news-briefing

# Verify installation
ls ~/.openclaw/skills/daily-news-briefing/

Option B: From Local Directory

bash

# Copy skill to OpenClaw skills directory
cp -r /path/to/daily-news-briefing ~/.openclaw/skills/

# Verify structure
ls ~/.openclaw/skills/daily-news-briefing/scripts/

Step 2: Configure Environment Variables (Optional)#

Note: The skill has a built-in fallback to DuckDuckGo if no Baidu API key is configured. Baidu API provides better structured data, but DuckDuckGo works perfectly fine without any API keys.

With Baidu API (Recommended for Better Results):

Create a configuration file with your Baidu Search API key:

Option A: System-wide (Recommended for servers)

bash

# Create config file
sudo nano /etc/profile.d/daily-news-briefing.sh

# Add your configuration
export BAIDU_API_KEY="bce-v3/ALTAK-your-api-key-here"
export NEWS_TARGET_USER="9C12E02D9038B14FCEDCE1B69AAEAB3F"  # QQ user ID
export NEWS_CHANNEL="qqbot"  # qqbot, telegram, discord

# Reload configuration
source /etc/profile.d/daily-news-briefing.sh

Option B: User-specific

bash

# Add to ~/.bashrc or ~/.zshrc
echo 'export BAIDU_API_KEY="your-api-key"' >> ~/.bashrc
echo 'export NEWS_TARGET_USER="target-user-id"' >> ~/.bashrc
source ~/.bashrc

Without API Key (Uses DuckDuckGo):

The skill will automatically use DuckDuckGo web search if no Baidu API key is configured:

bash

# Just set target user and channel - that's it!
export NEWS_TARGET_USER="your-qq-user-id"
export NEWS_CHANNEL="qqbot"

Compare Search Methods:

FeatureBaidu APIDuckDuckGoAPI Key Required✅ Yes (75 chars)❌ NoResult Quality🏆 Better structured data👍 Good for most use casesRate Limits⚠️ API quota applies✅ NoneSetup Time~2 mins (get API key)0 mins (works out of box)Content PreviewFull article snippetsTitle + URL only

Recommendation: Start with DuckDuckGo to test quickly, add Baidu API later if you need better content previews.

Step 3: Customize Delivery Settings (Optional)#

Edit the delivery script to match your preferences:

bash

nano ~/.openclaw/skills/daily-news-briefing/scripts/deliver-briefing.sh

Key settings to modify:

bash

# Change target user ID
TARGET_USER="your-qq-user-id"  # Line ~10

# Change delivery channel (--channel parameter)
--channel qqbot      # QQ Bot (default)
--channel telegram   # Telegram
--channel discord    # Discord

Step 4: Set Up Cron Job for Automated Delivery#

For daily delivery at 9:00 AM:

bash

# Edit crontab
crontab -e

Add these lines (adjust paths if needed):

cron

# Generate news at 9:00 AM
0 9 * * * source /etc/profile && cd ~/.openclaw/skills/daily-news-briefing/scripts && python3 generate-briefing.py >> /var/log/daily-news.log 2>&1

# Deliver at 9:01 AM
1 9 * * * source /etc/profile && bash ~/.openclaw/skills/daily-news-briefing/scripts/deliver-briefing.sh >> /var/log/news-delivery.log 2>&1

Custom delivery times:

TimeCron Entry (Generate + Deliver)Use Case7:00 AM0 7 * * * ...

1 7 * * * ...Early morning briefing7:30 AM30 7 * * * ...

31 7 * * * ...After system fully up (recommended)6:00 PM0 18 * * * ...

1 18 * * * ...Evening summary

Step 5: Test the Setup#

Test API Key:

bash

python3 -c "from generate_briefing import search_baidu; print(search_baidu('test', count=1))"

Generate News Manually:

bash

cd ~/.openclaw/skills/daily-news-briefing/scripts
python3 generate-briefing.py

Check log: tail -20 /var/log/daily-news.log

Test Delivery:

bash

bash deliver-briefing.sh

Check log: tail -30 /var/log/news-delivery.log

Verify Cron is Running:

bash

# Check cron service
systemctl status cron

# View scheduled jobs
crontab -l

# Check last execution
grep "Starting news generation" /var/log/daily-news.log | tail -1

Step 6: Customize News Content (Optional)#

Modify search queries in generate-briefing.py:

python

# Line ~85-90, customize keywords
china_tech = search_baidu('科技新闻 人工智能 芯片 AI 华为', count=3)
intl_tech = search_baidu('NVIDIA Broadcom Apple Microsoft AI', count=3)
china_finance = search_baidu('A 股 上证指数 港股 财经', count=3)
intl_finance = search_baidu('美股 纳斯达克 道琼斯 比特币', count=3)

Adjust article count: Change count=3 to 1-5 articles per category.

Customize AI commentary rules: See references/CONFIGURATION.md for pattern examples.

Components#

Scripts#

• generate-briefing.py: Main news generation script. Fetches from 4 categories:
  • China Tech News (AI, chips, Huawei, etc.)
  • International Tech News (NVIDIA, Apple, Microsoft, etc.)
  • China Financial Markets (A-shares, HK stocks)
  • International Finance (US stocks, Fed, crypto)
• deliver-briefing.sh: Delivery wrapper with multiple fallback strategies:
  • Method A: Direct OpenClaw CLI message send
  • Method B: Python delivery script
  • Method C: Create notification file
• news-deliver-direct.py: Alternative Python-based delivery with headline extraction

References#

See references/ for:

• CONFIGURATION.md: Detailed setup guide and customization options
• API_REFERENCE.md: Baidu API integration details and search query examples
• TEMPLATE_EXAMPLES.md: Sample briefing outputs and markdown templates

Customization#

Change Delivery Time#

Edit crontab lines (format: minute hour day month weekday):

cron

# 7:30 AM delivery
30 7 * * * ...
31 7 * * * ...

# 6:00 AM delivery
0 6 * * * ...
1 6 * * * ...

Modify Search Queries#

Edit queries in generate-briefing.py:

python

china_tech = search_baidu('科技新闻 人工智能 芯片 AI 华为', count=3)
intl_tech = search_baidu('NVIDIA Broadcom Apple Microsoft AI', count=3)

Add Custom Commentary Rules#

See references/API_REFERENCE.md for commentary rule patterns.

Troubleshooting#

No results from Baidu API:

• Verify BAIDU_API_KEY is set and valid (length > 20 chars)
• Check /var/log/daily-news.log for detailed errors

Delivery fails:

• Check /var/log/news-delivery.log
• Verify OpenClaw CLI is installed: openclaw --version
• Confirm target user ID is correct in script config

Wrong headlines extracted:

• Script supports both Chinese and English headers
• Ensure markdown file has proper structure with ## 🖥️ and ## 📈 sections

Multi-Channel Delivery Configuration#

Supported Channels#

The skill supports any OpenClaw-configured messaging channel:

ChannelTarget FormatExample ValueNotesQQBotc2c:USERID``c2c:9C12E02D...Default, for private QQ messagesTelegram@username or chat ID@mychannel or 123456789Bot must be added to channelDiscordguild-id/channel-id``1234567890/9876543210Bot needs permissionsSlack#channel-name or @user``#general or U123456Slack app configured

Configure Channel in Script#

Method 1: Environment Variable

bash

export NEWS_CHANNEL="telegram"  # qqbot, telegram, discord, slack
export NEWS_TARGET_USER="@mychannel"  # Channel-specific target

Method 2: Edit deliver-briefing.sh

Find line ~50 and modify:

bash

--channel qqbot      # Change to your preferred channel
-t "qqbot:c2c:${TARGET_USER}"  # Update target format per channel

Channel-Specific Setup#

For Telegram:

1. Create a bot via @BotFather
2. Add bot to channel/group
3. Get chat ID using openclaw status
4. Set NEWS_TARGET_USER="@channelname" or numeric ID

For Discord:

1. Create application in Discord Developer Portal
2. Invite bot to server with proper permissions
3. Use format: guild-id/channel-id
4. Test with: openclaw message send --channel discord -t "GUILD/CHANNEL" -m "Test"

For QQBot:

• Ensure QQ Bot is configured in OpenClaw
• Use c2c format for private messages
• Target user ID available from chat metadata

Output Format#

Generated briefing includes:

• Date and timestamp header
• 4 news categories with 3 articles each (customizable)
• AI-powered Jarvis commentary for each article
• Source attribution and read-more links
• File size: typically 10-15KB

Delivery message includes:

• Preview with first headlines from tech & finance sections
• <qqfile> attachment tag for full markdown report
• Footer with next delivery time notice

Logs & Monitoring#

Log FileLocationPurposeNews generation/var/log/daily-news.logSearch results, article counts, errorsDelivery attempts/var/log/news-delivery.logSend status, fallback methods used

View recent logs:

bash

# Today's news generation
tail -50 /var/log/daily-news.log

# Latest delivery result
grep "completed\|ERROR" /var/log/news-delivery.log | tail -10

Troubleshooting#

No results from Baidu API:

• Verify BAIDU_API_KEY is set and valid (length > 20 chars)
• Check /var/log/daily-news.log for "Found X results" messages
• Test: python3 -c "from generate_briefing import search_baidu; print(search_baidu('test', count=1))"
• Fallback: If Baidu API fails, script automatically uses DuckDuckGo (look for "falling back to DuckDuckGo" in logs)

No results from any source:

• Check network connectivity: ping -c 2 duckduckgo.com
• Verify firewall isn't blocking outbound HTTP/HTTPS
• Test DuckDuckGo directly: curl -s "https://html.duckduckgo.com/html/?q=test"
• Review /var/log/daily-news.log for detailed error messages

Delivery fails:

• Check /var/log/news-delivery.log for specific error
• Verify OpenClaw CLI is installed: openclaw --version
• Confirm target user/channel ID is correct in script config
• Test manual delivery: bash deliver-briefing.sh

Wrong headlines extracted:

• Script supports both Chinese ( ## 🖥️ 中国科技新闻) and English headers
• Ensure markdown file has proper structure with section markers
• Check log shows file size > 1KB (not empty fallback)
• DuckDuckGo results may have longer titles - this is normal

Cron job not running:

• Verify cron service: systemctl status cron
• Check crontab entries: crontab -l
• Test cron timing: wait for scheduled time or adjust to test immediately
• Check log rotation: old logs may be archived

Check which search method is being used:

bash

# View log file
grep "Search method used" /var/log/daily-news.log | tail -1

# Or check footer of generated news file
tail -5 ~/daily-news-$(date +%Y%m%d).md

Expected output: 📊 Search method used: baidu or 📊 Search method used: duckduckgo

Notes#

• Script auto-generates filename with current date: daily-news-YYYYMMDD.md
• Logs stored in /var/log/daily-news.log and /var/log/news-delivery.log
• Fallback content generated if search returns no results (graceful degradation)
• Supports multiple messaging channels via OpenClaw CLI
• All credentials managed via environment variables (no hardcoded secrets)

---

Version: 1.0

Last Updated: 2026-03-28

Maintained by: Jarvis AI Assistant