⚠️ December 2025 Urgent Notice

Due to a surge in Fork numbers causing excessive load on GitHub servers, GitHub Actions and GitHub Pages deployments are currently restricted. Please read the following instructions carefully to ensure successful deployment.

This is currently the most stable solution, free from GitHub restrictions. Data is stored locally and won't be affected by GitHub policy changes.

• 👉 Jump to Docker Deployment Tutorial

To reduce pressure on GitHub servers, please DO NOT directly click the "Fork" button!

Please use the "Use this template" feature instead of Fork:

1. Click the green [Use this template] button in the top right corner of the original repository page.
2. Select "Create a new repository".

Why do this?

• ❌ Fork: Copies complete history records. Many forks running simultaneously will trigger GitHub risk control.
• ✅ Use this template: Creates a completely new independent repository without historical baggage, more server-friendly.

The new version will use Cloudflare R2 to store news data, ensuring data persistence.

⚠️ Configuration Prerequisites:

According to Cloudflare platform rules, activating R2 requires binding a payment method.

• Purpose: Identity verification only (Verify Only), no charges will be incurred.
• Payment: Supports credit cards or PayPal (China region).
• Usage: R2's free tier is sufficient to cover this project's daily operation, no payment required.

Future Plans:

• Exploring new approach: keep Actions for fetching and pushing, but no longer save data to repository, use external storage instead.

⚠️ Reading Note: Given that the above plans mean Fork deployment mode may return in a new form in the future, and the workload to fully revise documentation is massive, we have temporarily retained the old descriptions.

At the current stage, if "Fork" related expressions still appear in subsequent tutorials, please ignore them or understand them as "Use this template".

👉 Click here to view TrendRadar's latest official documentation

Thanks to GitHub for providing free infrastructure, which is the biggest prerequisite for this project to run conveniently with one-click fork.

This project uses the API from newsnow to fetch multi-platform data. Special thanks to the author for providing this service.

After communication, the author indicated no concerns about server pressure, but this is based on their goodwill and trust. Please everyone:

• Visit the newsnow project and give it a star
• When deploying with Docker, please control the frequency reasonably and avoid being overly greedy

Thanks to the following platforms and individuals for recommendations (in chronological order)

• Appinn (小众软件) - Open source software recommendation platform
• LinuxDo Community - Tech enthusiasts community
• Ruan Yifeng's Weekly - Influential tech weekly in Chinese tech circle

Thanks to financial supporters. Your generosity has transformed into snacks and drinks beside my keyboard, accompanying every iteration of this project

"One-yuan appreciation" has been suspended. If you still want to support the author, please visit the official account article and click "Like Author" at the bottom.

🎉 Added Slack Push Support

1. Team Collaboration Push Channel

   • Supports Slack Incoming Webhooks (globally popular team collaboration tool)
   • Centralized message management, suitable for team-shared trending news
   • Supports mrkdwn format (bold, links, etc.)

2. Multiple Deployment Methods

   • GitHub Actions: Configure SLACK_WEBHOOK_URL Secret
   • Docker: Environment variable SLACK_WEBHOOK_URL
   • Local: config/config.yaml configuration file

📖 Detailed Configuration Tutorial: Quick Start - Slack Push

• Optimized the one-click installation experience for setup-windows.bat and setup-windows-en.bat

🔧 Upgrade Instructions:

• GitHub Fork Users: Update main.py, config/config.yaml, .github/workflows/crawler.yml

🎉 Added Bark Push Support

1. iOS Exclusive Push Channel

   • Supports Bark push (based on APNs, iOS platform)
   • Free, open-source, clean, efficient, ad-free
   • Supports both official server and self-hosted server

2. Multiple Deployment Methods

   • GitHub Actions: Configure BARK_URL Secret
   • Docker: Environment variable BARK_URL
   • Local: config/config.yaml configuration file

📖 Detailed Configuration Tutorial: Quick Start - Bark Push

🐛 Bug Fix

• Fixed issue where ntfy_server_url in config.yaml was ignored (#345)

🔧 Upgrade Instructions:

• GitHub Fork Users: Update main.py, config/config.yaml, .github/workflows/crawler.yml

🎯 New Advanced Customization Features

1. Keyword Sorting Priority Configuration

   • Two sorting strategies: Popularity first vs Config order first
   • For different use cases: Hot topic tracking or personalized focus

2. Display Count Precise Control

   • Global config: Unified limit for all keywords
   • Individual config: Use @number syntax to set specific limits
   • Effectively control push length, highlight key content

📖 Detailed Tutorial: Keyword Configuration - Advanced Settings

🔧 Upgrade Instructions:

• GitHub Fork Users: Update main.py, config/config.yaml

• Fixed data anomaly crash issue: Resolved 'float' object has no attribute 'lower' error encountered by some users in GitHub Actions environment
• Added dual protection mechanism: Filter invalid titles (None, float, empty strings) at data acquisition stage, with type checking at function call sites
• Enhanced system stability to ensure normal operation even when data sources return abnormal formats

Upgrade Instructions (GitHub Fork Users):

• Required update: main.py
• Recommended: Use minor version upgrade method - copy and replace the file above

MCP Module Update:

• Fix issue where today's news query may return articles from past dates

• Added Personal WeChat Push Support: WeWork application can push to personal WeChat without installing WeWork APP
• Supports two message formats: markdown (WeWork group bot) and text (personal WeChat app)
• Added WEWORK_MSG_TYPE environment variable configuration, supporting GitHub Actions, Docker, docker compose and other deployment methods
• text mode automatically strips Markdown syntax for clean plain text push
• See "Personal WeChat Push" configuration in Quick Start

Upgrade Instructions (GitHub Fork Users):

• Required updates: main.py, config/config.yaml
• Optional update: .github/workflows/crawler.yml (if using GitHub Actions)
• Recommended: Use minor version upgrade method - copy and replace the files above

• Fixed email sending SSL/TLS port configuration logic error
• Optimized email service providers (QQ/163/126) to default use port 465 (SSL)
• Added Docker environment variable support: Core config items (enable_crawler, report_mode, push_window, etc.) support override via environment variables, solving config file modification issues for NAS users (see 🐳 Docker Deployment chapter)

MCP Module Update:

• Fixed date query parameter passing error
• Unified time parameter format for all tools

• Solved Feishu error due to overly long push content, implemented batch pushing

• Expanded ntfy error message display range

• Fixed ntfy push encoding issue

Major Update - AI Analysis Feature Launched 🤖

• Core Features:

  • New MCP (Model Context Protocol) based AI analysis server
  • 13 smart analysis tools: basic query, smart search, advanced analysis, system management
  • Natural language interaction: Query and analyze news data through conversation
  • Multi-client support: Claude Desktop, Cherry Studio, Cursor, Cline, etc.

• Analysis Capabilities:

  • Topic trend analysis (popularity tracking, lifecycle, viral detection, trend prediction)
  • Data insights (platform comparison, activity stats, keyword co-occurrence)
  • Sentiment analysis, similar news finding, smart summary generation
  • Historical related news search, multi-mode search

• Update Note:

  • This is an independent AI analysis feature, does not affect existing push functionality
  • Optional use, no need to upgrade existing deployment

• Updates:

  • Fixed ntfy push encoding issue + 1
  • Fixed push time window judgment issue

• Upgrade Note:

  • Recommended minor version upgrade

Thanks to nidaye996 for discovering the UX issue

• Updates:

  • Refactored "Silent Push Mode" naming to "Push Time Window Control", improving feature comprehension
  • Clarified push time window as optional additional feature, can work with three push modes
  • Improved comments and documentation, making feature positioning clearer

• Upgrade Note:

  • This is just refactoring, upgrade optional

• Updates:

  • Fixed ntfy push encoding issue
  • Fixed missing config file issue
  • Optimized ntfy push effect
  • Added GitHub Pages image segmented export feature

• Upgrade Note:

  • Recommend major version update

Added ntfy Push Notification

• Core Features:

  • Supports ntfy.sh public service and self-hosted servers

• Use Cases:

  • Suitable for privacy-conscious users (supports self-hosting)
  • Cross-platform push (iOS, Android, Desktop, Web)
  • No account registration needed (public servers)
  • Open-source and free (MIT License)

• Upgrade Note:

  • Recommend major version update

• Fixed email notification config check being missed (#88)

Fix Description:

• Solved the issue where system still prompted "No webhook configured" even with correct email notification setup

• Added email push feature, supports sending trending news reports to email
• Smart SMTP Recognition: Auto-detects Gmail, QQ Mail, Outlook, NetEase Mail and 10+ email service providers
• Beautiful HTML Format: Email content uses same HTML format as web version, well-formatted, mobile-adapted
• Batch Sending Support: Supports multiple recipients, separated by commas
• Custom SMTP: Can customize SMTP server and port
• Fixed Docker build network connection issue

Usage Notes:

• Use cases: Suitable for users needing email archiving, team sharing, scheduled reports
• Supported emails: Gmail, QQ Mail, Outlook/Hotmail, 163/126 Mail, Sina Mail, Sohu Mail, etc.

Upgrade Note:

• This update has many changes, if upgrading, recommend major version upgrade

• Added one-click save news as image feature, easily share trending topics you care about

Usage Notes:

• Use case: After enabling web version feature (GitHub Pages)
• How to use: Open webpage on phone or PC, click "Save as Image" button at top
• Actual effect: System auto-creates beautiful image of current news report, saves to phone album or desktop
• Sharing convenience: Directly send this image to friends, Moments, or work groups, letting others see your discovered important info

• Solved DingTalk push capacity limit causing news push failure (using batch push)

• Fixed Docker unable to run properly on certain architectures
• Officially released official Docker image wantcat/trendradar, supports multi-architecture
• Optimized Docker deployment process, can use quickly without local build

Core Improvements:

• Push Logic Optimization: Changed from "push every execution" to "controllable push within time window"
• Time Window Control: Can set push time range, avoid non-work hour disturbance
• Push Frequency Options: Supports single push or multiple pushes within time window

Upgrade Note:

• This feature is disabled by default, need to manually enable push time window control in config.yaml
• Upgrade requires updating both main.py and config.yaml files

• This version is not a bug fix, but an important reminder
• Please keep webhooks properly, do not make public, do not make public, do not make public
• If you deployed this project on GitHub via fork, please put webhooks in GitHub Secret, not config.yaml
• If you already exposed webhooks or put them in config.yaml, suggest deleting and regenerating

• Optimized GitHub Pages web version effect, convenient for mobile use

• Refactored code
• Solved version number easily being missed for modification

Fixed Issues:

1. Docker shell script line ending as CRLF causing execution exception issue
2. frequency_words.txt being empty causing news sending also empty logic issue

• After fix, when you choose frequency_words.txt empty, will push all news, but limited by message push size, please adjust as follows
  • Option 1: Turn off mobile push, only choose GitHub Pages deployment (this is the way to get most complete info, will re-sort all platform trending by your custom trending algorithm)
  • Option 2: Reduce push platforms, prioritize WeWork or Telegram, these two pushes I made batch push feature (because batch push affects push experience, and only these two platforms give very little push capacity, so had to make batch push feature, but at least can ensure complete info)
  • Option 3: Can combine with Option 2, mode choose current or incremental can effectively reduce one-time push content

Major Refactoring:

• Config management refactoring: All configs now managed through config/config.yaml file (main.py I still didn't split, convenient for you to copy and upgrade)
• Run mode upgrade: Supports three modes - daily (daily summary), current (current rankings), incremental (incremental monitoring)
• Docker support: Complete Docker deployment solution, supports containerized operation

Config File Description:

• config/config.yaml - Main config file (application settings, crawler config, notification config, platform config, etc.)
• config/frequency_words.txt - Keyword config (monitoring vocabulary settings)

New Feature: Added incremental push (configure FOCUS_NEW_ONLY at top of main.py), this switch only cares about new topics not sustained heat, only sends notification when new content appears.

Fixed Issue: Under certain circumstances, some news containing special symbols caused occasional formatting exceptions.

WeWork and Telegram push messages have length limits, I adopted splitting messages for pushing. Development docs see WeWork and Telegram

Before this version, not only main.py needs copy replacement, crawler.yml also needs you to copy replacement https://github.com/sansan0/TrendRadar/blob/master/.github/workflows/crawler.yml

Thanks to Claude Research for organizing various platform APIs, helping me quickly complete platform adaptation (although code is more redundant~

1. Supports Telegram, WeWork, DingTalk push channels, supports multi-channel config and simultaneous push

200 stars⭐ reached, continue celebrating with everyone~

1. Important update, added weight, news you see now is hottest most concerned appearing at top
2. Updated documentation usage, because recently updated many features, and previous usage docs I was lazy wrote simple (see ⚙️ frequency_words.txt complete configuration tutorial below)

1. Added project new version update reminder, default on, if want to turn off, can change "FEISHU_SHOW_VERSION_UPDATE": True to False in main.py

1. Removed compatibility code, students who forked before, directly copying code will show exception on same day (will recover normal next day)
2. Feishu and html bottom added new news display

100 stars⭐ reached, writing small feature to celebrate

frequency_words.txt file added required word feature, using + sign

1. Required word syntax as follows: Tang Monk or Pig must both appear in title, will be included in push news

+Tang Monk
+Pig

2. Filter word priority higher: If title filter word matches Tang Monk reciting sutras, then even if required word has Tang Monk, also not display

+Tang Monk
!Tang Monk reciting sutras

1. Webpage and Feishu messages support phone directly jumping to detailed news
2. Optimized display effect + 1

1. Feishu message display effect optimized

Configuration Location: platforms section in config/config.yaml

This project's news data comes from newsnow. You can click the website, click [More], to see if there are platforms you want.

For specific additions, visit project source code, based on the file names there, modify the platforms configuration in config/config.yaml file:

platforms:
  - id: "toutiao"
    name: "Toutiao"
  - id: "baidu"
    name: "Baidu Hot Search"
  - id: "wallstreetcn-hot"
    name: "Wallstreetcn"
  # Add more platforms...

💡 Shortcut: If you don't know how to read source code, you can copy from others' organized Platform Configuration Summary

⚠️ Note: More platforms is not always better, suggest choosing 10-15 core platforms. Too many platforms will cause information overload and actually reduce user experience.

Huawei
OPPO
Apple

Effect: News containing any one of these words will be captured

Huawei
OPPO
+phone

Effect: Must include both normal word and required word to be captured

Apple
Huawei
!fruit
!price

Effect: News containing filter words will be excluded, even if it contains keywords

Tesla
Musk
@5

Effect: Limit maximum news count for this keyword group

Priority: @number > Global config > Unlimited

[GLOBAL_FILTER]
advertisement
promotion
marketing
shocking
clickbait

[WORD_GROUPS]
technology
AI

Huawei
HarmonyOS
!car

Effect: Filters news containing specified words under any circumstances, with highest priority

Use Cases:

• Filter low-quality content: shocking, clickbait, breaking news, etc.
• Filter marketing content: advertisement, promotion, sponsorship, etc.
• Filter specific topics: entertainment, gossip (based on needs)

Filter Priority: Global Filter > Group Filter(!) > Group Matching

Region Markers:

• [GLOBAL_FILTER]: Global filter region, words are filtered under any circumstances
• [WORD_GROUPS]: Keyword groups region, maintains existing syntax (!, +, @)
• If no region markers are used, all content is treated as keyword groups (backward compatible)

Matching Examples:

[GLOBAL_FILTER]
advertisement

[WORD_GROUPS]
technology
AI

• ❌ "Advertisement: Latest tech product launch" ← Contains global filter word "advertisement", rejected
• ✅ "Tech company launches new AI product" ← No global filter words, matches "technology" group
• ✅ "AI technology breakthrough draws attention" ← No global filter words, matches "AI" in "technology" group

Important Notes:

• Use global filter words carefully to avoid over-filtering and missing valuable content
• Recommended to keep global filter words under 5-15
• For group-specific filtering, prioritize using group filter words (! prefix)

Core Rule: Use empty lines to separate different groups, each group is independently counted

iPhone
Huawei
OPPO
+launch

A-shares
Shanghai Index
Shenzhen Index
+fluctuation
!prediction

World Cup
Euro Cup
Asian Cup
+match

Group 1 - Phone Launches:

• Keywords: iPhone, Huawei, OPPO
• Required: launch
• Effect: Must include phone brand name and "launch"

Matching Examples:

• ✅ "iPhone 15 officially launched with pricing" ← Has "iPhone" + "launch"
• ✅ "Huawei Mate60 series launch livestream" ← Has "Huawei" + "launch"
• ✅ "OPPO Find X7 launch date confirmed" ← Has "OPPO" + "launch"
• ❌ "iPhone sales hit record high" ← Has "iPhone" but missing "launch"

Group 2 - Stock Market:

• Keywords: A-shares, Shanghai Index, Shenzhen Index
• Required: fluctuation
• Filter: prediction
• Effect: Include stock-related words and "fluctuation", but exclude "prediction"

Matching Examples:

• ✅ "A-shares major fluctuation analysis today" ← Has "A-shares" + "fluctuation"
• ✅ "Shanghai Index fluctuation reasons explained" ← Has "Shanghai Index" + "fluctuation"
• ❌ "Experts predict A-shares fluctuation trends" ← Has "A-shares" + "fluctuation" but contains "prediction"
• ❌ "A-shares trading volume hits new high" ← Has "A-shares" but missing "fluctuation"

Group 3 - Football Events:

• Keywords: World Cup, Euro Cup, Asian Cup
• Required: match
• Effect: Must include cup name and "match"

Matching Examples:

• ✅ "World Cup group stage match results" ← Has "World Cup" + "match"
• ✅ "Euro Cup final match time" ← Has "Euro Cup" + "match"
• ❌ "World Cup tickets on sale" ← Has "World Cup" but missing "match"

# Step 1: Start with broad keywords for testing
Artificial Intelligence
AI
ChatGPT

# Step 2: After finding mismatches, add required words
Artificial Intelligence
AI
ChatGPT
+technology

# Step 3: After finding noise, add filter words
Artificial Intelligence
AI
ChatGPT
+technology
!advertisement
!training

❌ Not Recommended: Too many words in one group

Huawei
OPPO
Apple
Samsung
vivo
OnePlus
Meizu
+phone
+launch
+sales
!fake
!repair
!second-hand

Recommended: Split into precise groups

Huawei
OPPO
+new product

Apple
Samsung
+launch

phone
sales
+market

Config Location: config/config.yaml

report:
  sort_by_position_first: false  # Sorting priority config

Example: Config order A, B, C, news count A(3), B(10), C(5)

• false: B(10) → C(5) → A(3)
• true: A(3) → B(10) → C(5)

report:
  max_news_per_keyword: 10  # Max 10 per keyword (0=unlimited)

Docker Environment Variables:

SORT_BY_POSITION_FIRST=true
MAX_NEWS_PER_KEYWORD=10

Combined Example:

# config.yaml
report:
  sort_by_position_first: true   # Config order priority
  max_news_per_keyword: 10       # Global default max 10 per keyword

# frequency_words.txt
Tesla
Musk
@20              # Key focus, show 20 (override global)

Huawei           # Use global config, show 10

BYD
@5               # Limit to 5

Final Effect: Display in config order: Tesla(20) → Huawei(10) → BYD(5)

Configuration Location: report.mode in config/config.yaml

report:
  mode: "daily"  # Options: "daily" | "incremental" | "current"

Docker Environment Variable: REPORT_MODE=incremental

Assume you monitor "Apple" keyword, execute once per hour:

Explanation:

• daily: Cumulative display of all news of the day (A, B, C all retained)
• current: Display current ranking news (ranking changed, News D on list, News A off list)
• incremental: Only push newly appeared news (avoid duplicate disturbance)

💡 Encountered this problem? 👉 "Execute once per hour, news output in first execution still appears in next hour execution"

• Reason: You might have selected daily (Daily Summary) or current (Current Rankings) mode
• Solution: Change to incremental (Incremental Monitor) mode, only push new content

Users who selected incremental (Incremental Monitor) mode, please note:

📌 Incremental mode only pushes when there are new matching news

If you haven't received push notifications for a long time, it may be because:

1. No new hot topics matching your keywords in current time period
2. Keyword configuration is too strict or too broad
3. Too few monitoring platforms

Solutions:

• Solution 1: 👉 Optimize Keyword Configuration - Adjust keyword precision, add or modify monitoring keywords
• Solution 2: Switch push mode - Change to current or daily mode for scheduled push notifications
• Solution 3: 👉 Add More Platforms - Add more news platforms to expand information sources

Configuration Location: weight section in config/config.yaml

weight:
  rank_weight: 0.6       # Ranking weight
  frequency_weight: 0.3  # Frequency weight
  hotness_weight: 0.1    # Hotness weight

Current default configuration is balanced.

Real-Time Trending Type:

weight:
  rank_weight: 0.8    # Mainly focus on ranking
  frequency_weight: 0.1  # Less concern about continuity
  hotness_weight: 0.1

Target Users: Content creators, marketers, users wanting to quickly understand current hot topics

In-Depth Topic Type:

weight:
  rank_weight: 0.4    # Moderate ranking focus
  frequency_weight: 0.5  # Emphasize sustained heat within the day
  hotness_weight: 0.1

Target Users: Investors, researchers, journalists, users needing deep trend analysis

1. Three numbers must sum to 1.0
2. Increase what's important: Increase rank_weight for rankings, frequency_weight for continuity
3. Suggest adjusting 0.1-0.2 at a time, observe effects

Core idea: Users pursuing speed and timeliness increase ranking weight, users pursuing depth and stability increase frequency weight.

📊 Trending Keywords Stats

🔥 [1/3] AI ChatGPT : 2 items

1. [Baidu Hot] 🆕 ChatGPT-5 officially launched [1] - 09:15 (1 time)

2. [Toutiao] AI chip concept stocks surge [3] - [08:30 ~ 10:45] (3 times)

━━━━━━━━━━━━━━━━━━━

📈 [2/3] BYD Tesla : 2 items

1. [Weibo] 🆕 BYD monthly sales break record [2] - 10:20 (1 time)

2. [Douyin] Tesla price reduction promotion [4] - [07:45 ~ 09:15] (2 times)

━━━━━━━━━━━━━━━━━━━

📌 [3/3] A-shares Stock Market : 1 item

1. [Wallstreetcn] A-shares midday review [5] - [11:30 ~ 12:00] (2 times)

🆕 New Trending News (Total 2 items)

Baidu Hot (1 item):

1. ChatGPT-5 officially launched [1]

Weibo (1 item):

1. BYD monthly sales break record [2]

Updated: 2025-01-15 12:30:15

Image Description:

TrendRadar provides two independent Docker images, deploy according to your needs:

💡 Recommendations:

• Only need push functionality: Deploy wantcat/trendradar image only
• Need AI analysis: Deploy both images

1. Create Project Directory and Config:

   Method 1-A: Using git clone (Recommended, Simplest)

   # Clone project to local
   git clone https://github.com/sansan0/TrendRadar.git
   cd TrendRadar
   

   Method 1-B: Using wget to download config files

   # Create directory structure
   mkdir -p trendradar/{config,docker}
   cd trendradar
   
   # Download config file templates
   wget https://raw.githubusercontent.com/sansan0/TrendRadar/master/config/config.yaml -P config/
   wget https://raw.githubusercontent.com/sansan0/TrendRadar/master/config/frequency_words.txt -P config/
   
   # Download docker compose config
   wget https://raw.githubusercontent.com/sansan0/TrendRadar/master/docker/.env -P docker/
   wget https://raw.githubusercontent.com/sansan0/TrendRadar/master/docker/docker compose.yml -P docker/
   

   💡 Note: Key directory structure required for Docker deployment:

current directory/
├── config/
│   ├── config.yaml
│   └── frequency_words.txt
└── docker/
    ├── .env
    └── docker compose.yml

2. Config File Description:

   • config/config.yaml - Application main config (report mode, push settings, etc.)
   • config/frequency_words.txt - Keyword config (set your interested trending keywords)
   • .env - Environment variable config (webhook URLs and scheduled tasks)

   ⚙️ Environment Variable Override Mechanism (v3.0.5+)

   If you encounter config.yaml modifications not taking effect in NAS or other Docker environments, you can directly override configs via environment variables:

   Environment Variable	Corresponding Config	Example Value	Description
   ENABLE_CRAWLER	crawler.enable_crawler	true / false	Enable crawler
   ENABLE_NOTIFICATION	notification.enable_notification	true / false	Enable notification
   REPORT_MODE	report.mode	daily / incremental / current	Report mode
   MAX_ACCOUNTS_PER_CHANNEL	notification.max_accounts_per_channel	3	Maximum accounts per channel
   PUSH_WINDOW_ENABLED	notification.push_window.enabled	true / false	Push time window switch
   PUSH_WINDOW_START	notification.push_window.time_range.start	08:00	Push start time
   PUSH_WINDOW_END	notification.push_window.time_range.end	22:00	Push end time
   ENABLE_WEBSERVER	-	true / false	Auto-start web server
   WEBSERVER_PORT	-	8080	Web server port (default 8080)
   FEISHU_WEBHOOK_URL	notification.webhooks.feishu_url	https://...	Feishu Webhook (supports multi-account, use ; separator)

   Config Priority: Environment Variables > config.yaml

   Usage Method:

   • Modify .env file, uncomment and fill in needed configs
   • Or add directly in NAS/Synology Docker management interface's "Environment Variables"
   • Restart container to take effect: docker compose up -d

3. Start Service:

   Option A: Start All Services (Push + AI Analysis)

   # Pull latest images
   docker compose pull
   
   # Start all services (trend-radar + trend-radar-mcp)
   docker compose up -d
   

   Option B: Start News Push Service Only

   # Start trend-radar only (scheduled crawling and push)
   docker compose pull trend-radar
   docker compose up -d trend-radar
   

   Option C: Start MCP AI Analysis Service Only

   # Start trend-radar-mcp only (AI analysis interface)
   docker compose pull trend-radar-mcp
   docker compose up -d trend-radar-mcp
   

   💡 Tips:

   • Most users only need to start trend-radar for news push functionality
   • Only start trend-radar-mcp when using Claude/ChatGPT for AI dialogue analysis
   • Both services are independent and can be flexibly combined

4. Check Running Status:

   # View news push service logs
   docker logs -f trend-radar
   
   # View MCP AI analysis service logs
   docker logs -f trend-radar-mcp
   
   # View all container status
   docker ps | grep trend-radar
   
   # Stop specific service
   docker compose stop trend-radar      # Stop push service
   docker compose stop trend-radar-mcp  # Stop MCP service
   

If you need custom code modifications or build your own image:

# Clone project
git clone https://github.com/sansan0/TrendRadar.git
cd TrendRadar

# Modify config files
vim config/config.yaml
vim config/frequency_words.txt

# Use build version docker compose
cd docker
cp docker compose-build.yml docker compose.yml

Build and Start Services:

# Option A: Build and start all services
docker compose build
docker compose up -d

# Option B: Build and start news push service only
docker compose build trend-radar
docker compose up -d trend-radar

# Option C: Build and start MCP AI analysis service only
docker compose build trend-radar-mcp
docker compose up -d trend-radar-mcp

💡 Architecture Parameter Notes:

• Default builds amd64 architecture images (suitable for most x86_64 servers)
• To build arm64 architecture (Apple Silicon, Raspberry Pi, etc.), set environment variable:

  export DOCKER_ARCH=arm64
  docker compose build
  

# Method 1: Manual update (Crawler + MCP images)
docker pull wantcat/trendradar:latest
docker pull wantcat/trendradar-mcp:latest
docker compose down
docker compose up -d

# Method 2: Using docker compose update
docker compose pull
docker compose up -d

Available Images:

# View running status
docker exec -it trend-radar python manage.py status

# Manually execute crawler once
docker exec -it trend-radar python manage.py run

# View real-time logs
docker exec -it trend-radar python manage.py logs

# Display current config
docker exec -it trend-radar python manage.py config

# Display output files
docker exec -it trend-radar python manage.py files

# Web server management (for browser access to generated reports)
docker exec -it trend-radar python manage.py start_webserver   # Start web server
docker exec -it trend-radar python manage.py stop_webserver    # Stop web server
docker exec -it trend-radar python manage.py webserver_status  # Check web server status

# View help info
docker exec -it trend-radar python manage.py help

# Restart container
docker restart trend-radar

# Stop container
docker stop trend-radar

# Remove container (keep data)
docker rm trend-radar

💡 Web Server Notes:

• After starting, access latest report at http://localhost:8080
• Access historical reports via directory navigation (e.g., http://localhost:8080/2025年xx月xx日/)
• Port can be configured in .env file with WEBSERVER_PORT parameter
• Auto-start: Set ENABLE_WEBSERVER=true in .env
• Security: Static files only, limited to output directory, localhost binding only

Generated reports and data are saved in ./output directory by default. Data persists even if container is restarted or removed.

📊 Web Report Access Paths:

TrendRadar generates daily summary HTML reports to two locations simultaneously:

Local Access Examples:

# Method 1: Via Web Server (recommended, Docker environment)
# 1. Start web server
docker exec -it trend-radar python manage.py start_webserver
# 2. Access in browser
http://localhost:8080                           # Access latest report (default index.html)
http://localhost:8080/2025年xx月xx日/            # Access reports for specific date
http://localhost:8080/2025年xx月xx日/html/       # Browse all HTML files for that date

# Method 2: Direct file access (local environment)
open ./output/index.html             # macOS
start ./output/index.html            # Windows
xdg-open ./output/index.html         # Linux

# Method 3: Access historical archives
open ./output/2025年xx月xx日/html/当日汇总.html

Why two index.html files?

• output/index.html: Docker Volume mounted to host, can be opened locally
• index.html: Pushed to repository by GitHub Actions, auto-deployed by GitHub Pages

💡 Tip: Both files have identical content, choose either one to access.

# Check container status
docker inspect trend-radar

# View container logs
docker logs --tail 100 trend-radar

# Enter container for debugging
docker exec -it trend-radar /bin/bash

# Verify config files
docker exec -it trend-radar ls -la /app/config/

If you need to use AI analysis features, you can deploy the standalone MCP service container.

Architecture Description:

Quick Start:

Use docker compose to start both news push and MCP services:

# Download latest docker compose.yml (includes MCP service config)
wget https://raw.githubusercontent.com/sansan0/TrendRadar/master/docker/docker compose.yml

# Start all services
docker compose up -d

# Check running status
docker ps | grep trend-radar

Start MCP Service Separately:

docker run -d --name trend-radar-mcp \
  -p 127.0.0.1:3333:3333 \
  -v ./config:/app/config:ro \
  -v ./output:/app/output:ro \
  -e TZ=Asia/Shanghai \
  wantcat/trendradar-mcp:latest

Verify Service:

# Check if MCP service is running properly
curl http://127.0.0.1:3333/mcp

# View MCP service logs
docker logs -f trend-radar-mcp

Configure in AI Clients:

After MCP service starts, configure in Claude Desktop, Cherry Studio, Cursor, etc.:

{
  "mcpServers": {
    "trendradar": {
      "url": "http://127.0.0.1:3333/mcp",
      "description": "TrendRadar News Trending Analysis"
    }
  }
}

💡 Tip: MCP service only listens on local port (127.0.0.1) for security. For remote access, configure reverse proxy and authentication yourself.

Configuration Location: report section in config/config.yaml

report:
  mode: "daily"                    # Push mode
  rank_threshold: 5                # Ranking highlight threshold
  sort_by_position_first: false    # Sorting priority
  max_news_per_keyword: 0          # Maximum display count per keyword
  reverse_content_order: false     # Content order configuration

Controls display order of two content sections in push messages and HTML reports:

Use Cases:

• false (default): Suitable for users focusing on keyword match results, view categorized stats first
• true: Suitable for users focusing on latest updates, prioritize viewing new trending topics

Docker Environment Variable:

REVERSE_CONTENT_ORDER=true

Example Scenario: Config order A, B, C, news count A(3), B(10), C(5)

Docker Environment Variables:

SORT_BY_POSITION_FIRST=true
MAX_NEWS_PER_KEYWORD=10

Configuration Location: notification.push_window section in config/config.yaml

notification:
  push_window:
    enabled: false                    # Whether to enable
    time_range:
      start: "20:00"                  # Start time (Beijing time)
      end: "22:00"                    # End time (Beijing time)
    once_per_day: true                # Push only once per day
    push_record_retention_days: 7     # Push record retention days

⚠️ GitHub Actions Users Note:

• GitHub Actions execution time is unstable, may have ±15 minutes deviation
• Time range should be at least 2 hours wide
• For precise timed push, recommend Docker deployment on personal server

PUSH_WINDOW_ENABLED=true
PUSH_WINDOW_START=09:00
PUSH_WINDOW_END=18:00
PUSH_WINDOW_ONCE_PER_DAY=false
PUSH_WINDOW_RETENTION_DAYS=7

Scenario: Push once between 8-10 PM daily

notification:
  push_window:
    enabled: true
    time_range:
      start: "20:00"
      end: "22:00"
    once_per_day: true
    push_record_retention_days: 7

Scenario: Push every hour during working hours

notification:
  push_window:
    enabled: true
    time_range:
      start: "09:00"
      end: "18:00"
    once_per_day: false
    push_record_retention_days: 7

Configuration Location: schedule section in .github/workflows/crawler.yml

on:
  schedule:
    - cron: "0 * * * *"  # Run every hour

Cron is a time-based job scheduler format, consisting of 5 parts: minute hour day month weekday

┌───────────── minute (0-59)
│ ┌───────────── hour (0-23)
│ │ ┌───────────── day (1-31)
│ │ │ ┌───────────── month (1-12)
│ │ │ │ ┌───────────── weekday (0-6, 0=Sunday)
│ │ │ │ │
* * * * *

⚠️ Time Zone Note: GitHub Actions uses UTC time, Beijing time needs to subtract 8 hours

• Want Beijing 8:00 AM run → Set UTC 0:00
• Want Beijing 8:00 PM run → Set UTC 12:00

⚠️ Frequency Limit: GitHub has a limit on Actions execution count per account

• Recommendation: Don't set intervals shorter than 30 minutes
• Reason: Too frequent may be considered abuse, facing account ban risk
• Reality: GitHub Actions execution time has inherent deviation, setting too precise is meaningless

1. Open your forked repository
2. Find .github/workflows/crawler.yml file
3. Click edit (pencil icon)
4. Modify the expression in cron: "0 * * * *"
5. Click "Commit changes" to save

Configuration Location: notification section in config/config.yaml

⚠️ Security Warning

GitHub Fork Users: DO NOT configure push information in config.yaml!

• Risk: config.yaml will be committed to public Git repositories. Configuring push information (Webhook URLs, Tokens, etc.) will expose sensitive data
• Recommended Methods:
  • GitHub Actions Users → Use GitHub Secrets environment variables
  • Docker Users → Use .env file configuration (.env is in .gitignore and won't be committed)
• Local Development Users: Can configure in config.yaml (ensure it won't be pushed to public repositories)

Configuration Location: GitHub Repo → Settings → Secrets and variables → Actions → Repository secrets

Basic Configuration Example:

# Multi-account quantity limit
MAX_ACCOUNTS_PER_CHANNEL=3

# Feishu multi-account (3 groups)
FEISHU_WEBHOOK_URL=https://hook1.feishu.cn/xxx;https://hook2.feishu.cn/yyy;https://hook3.feishu.cn/zzz

# DingTalk multi-account (2 groups)
DINGTALK_WEBHOOK_URL=https://oapi.dingtalk.com/xxx;https://oapi.dingtalk.com/yyy

# WeWork multi-account (2 groups)
WEWORK_WEBHOOK_URL=https://qyapi.weixin.qq.com/cgi-bin/webhook/send?key=xxx;https://qyapi.weixin.qq.com/cgi-bin/webhook/send?key=yyy

# Bark multi-account (2 devices)
BARK_URL=https://api.day.app/key1;https://api.day.app/key2

# Slack multi-account (2 channels)
SLACK_WEBHOOK_URL=https://hooks.slack.com/xxx;https://hooks.slack.com/yyy

Paired Configuration Examples (Telegram and ntfy):

# ✅ Correct: 2 tokens correspond to 2 chat_ids
TELEGRAM_BOT_TOKEN=123456:AAA-BBB;789012:CCC-DDD
TELEGRAM_CHAT_ID=-100111;-100222

# ❌ Incorrect: quantities don't match, push will be skipped
TELEGRAM_BOT_TOKEN=token1;token2;token3
TELEGRAM_CHAT_ID=id1;id2

# ✅ Correct: 3 topics, only the 2nd needs a token
NTFY_TOPIC=topic1;topic2;topic3
NTFY_TOKEN=;token_for_topic2;

# ✅ Correct: 2 topics both need tokens
NTFY_TOPIC=topic1;topic2
NTFY_TOKEN=token1;token2

# ❌ Incorrect: topic and token quantities don't match
NTFY_TOPIC=topic1;topic2
NTFY_TOKEN=token1;token2;token3

Configuration Location: docker/.env file in project root directory

Basic Configuration Example:

# Multi-account quantity limit
MAX_ACCOUNTS_PER_CHANNEL=3

# Feishu multi-account (3 groups)
FEISHU_WEBHOOK_URL=https://hook1.feishu.cn/xxx;https://hook2.feishu.cn/yyy;https://hook3.feishu.cn/zzz

# DingTalk multi-account (2 groups)
DINGTALK_WEBHOOK_URL=https://oapi.dingtalk.com/xxx;https://oapi.dingtalk.com/yyy

# WeWork multi-account (2 groups)
WEWORK_WEBHOOK_URL=https://qyapi.weixin.qq.com/cgi-bin/webhook/send?key=xxx;https://qyapi.weixin.qq.com/cgi-bin/webhook/send?key=yyy

# Bark multi-account (2 devices)
BARK_URL=https://api.day.app/key1;https://api.day.app/key2

# Slack multi-account (2 channels)
SLACK_WEBHOOK_URL=https://hooks.slack.com/xxx;https://hooks.slack.com/yyy

Paired Configuration Examples (Telegram and ntfy):

# ✅ Correct: 2 tokens correspond to 2 chat_ids
TELEGRAM_BOT_TOKEN=123456:AAA-BBB;789012:CCC-DDD
TELEGRAM_CHAT_ID=-100111;-100222

# ❌ Incorrect: quantities don't match, push will be skipped
TELEGRAM_BOT_TOKEN=token1;token2;token3
TELEGRAM_CHAT_ID=id1;id2

# ✅ Correct: 3 topics, only the 2nd needs a token
NTFY_TOPIC=topic1;topic2;topic3
NTFY_TOKEN=;token_for_topic2;

# ✅ Correct: 2 topics both need tokens
NTFY_TOPIC=topic1;topic2
NTFY_TOKEN=token1;token2

# ❌ Incorrect: topic and token quantities don't match
NTFY_TOPIC=topic1;topic2
NTFY_TOKEN=token1;token2;token3

1. Independent Push: Each account sends independently, one failure doesn't affect other accounts
2. Partial Success: As long as one account sends successfully, the overall result is considered successful
3. Log Differentiation: Multi-account logs show "Account 1", "Account 2", etc.
4. Batch Interval: Multi-account increases total send time (each account independently calculates batch interval)

notification:
  enable_notification: true
  max_accounts_per_channel: 3

  webhooks:
    feishu_url: "https://hook1.feishu.cn/xxx;https://hook2.feishu.cn/yyy"
    telegram_bot_token: "token1;token2"
    telegram_chat_id: "id1;id2"

💡 Tip: Actually not recommended to ask multiple questions at once. If your chosen AI model cannot even sequentially call as shown below, suggest switching models.

Edit Claude Desktop's MCP config file:

Windows: %APPDATA%\Claude\claude_desktop_config.json

Mac: ~/Library/Application Support/Claude/claude_desktop_config.json

Config Content:

{
  "mcpServers": {
    "trendradar": {
      "command": "uv",
      "args": [
        "--directory",
        "/path/to/TrendRadar",
        "run",
        "python",
        "-m",
        "mcp_server.server"
      ],
      "env": {},
      "disabled": false,
      "alwaysAllow": []
    }
  }
}

1. Start HTTP Service:

   # Windows
   start-http.bat
   
   # Mac/Linux
   ./start-http.sh
   

2. Configure Cursor:

   Project Level Config (Recommended): Create .cursor/mcp.json in project root:

   {
     "mcpServers": {
       "trendradar": {
         "url": "http://localhost:3333/mcp",
         "description": "TrendRadar News Trending Aggregation Analysis"
       }
     }
   }
   

   Global Config: Create ~/.cursor/mcp.json in user directory (same content)

3. Usage Steps:

   • Restart Cursor after saving config
   • Check connected tools in chat interface "Available Tools"
   • Start using: Search today's "AI" related news

Create .cursor/mcp.json:

{
  "mcpServers": {
    "trendradar": {
      "command": "uv",
      "args": [
        "--directory",
        "/path/to/TrendRadar",
        "run",
        "python",
        "-m",
        "mcp_server.server"
      ]
    }
  }
}

Add in Cline's MCP settings:

HTTP Mode:

{
  "trendradar": {
    "url": "http://localhost:3333/mcp",
    "type": "streamableHttp",
    "autoApprove": [],
    "disabled": false
  }
}

STDIO Mode (Recommended):

{
  "trendradar": {
    "command": "uv",
    "args": [
      "--directory",
      "/path/to/TrendRadar",
      "run",
      "python",
      "-m",
      "mcp_server.server"
    ],
    "type": "stdio",
    "disabled": false
  }
}

Edit ~/.continue/config.json:

{
  "experimental": {
    "modelContextProtocolServers": [
      {
        "transport": {
          "type": "stdio",
          "command": "uv",
          "args": [
            "--directory",
            "/path/to/TrendRadar",
            "run",
            "python",
            "-m",
            "mcp_server.server"
          ]
        }
      }
    ]
  }
}

Usage Examples:

Analyze recent 7 days "Tesla" popularity trend
Generate today's trending summary report
Search "Bitcoin" related news and analyze sentiment

# 1. Start HTTP service
# Windows: start-http.bat
# Mac/Linux: ./start-http.sh

# 2. Add MCP server
claude mcp add --transport http trendradar http://localhost:3333/mcp

# 3. Verify connection (ensure service started)
claude mcp list

# Query news
claude "Search today's Zhihu trending news, top 10"

# Trend analysis
claude "Analyze 'artificial intelligence' topic popularity trend for the past week"

# Data comparison
claude "Compare Zhihu and Weibo platform attention on 'Bitcoin'"

MCP Inspector is the official debug tool for testing MCP connections:

1. Start TrendRadar HTTP Service:

   # Windows
   start-http.bat
   
   # Mac/Linux
   ./start-http.sh
   

2. Start MCP Inspector:

   npx @modelcontextprotocol/inspector
   

3. Connect in Browser:

   • Visit: http://localhost:3333/mcp
   • Test "Ping Server" function to verify connection
   • Check "List Tools" returns 14 tools:
     • Date Parsing: resolve_date_range
     • Basic Query: get_latest_news, get_news_by_date, get_trending_topics
     • Smart Search: search_news, search_related_news_history
     • Advanced Analysis: analyze_topic_trend, analyze_data_insights, analyze_sentiment, find_similar_news, generate_summary_report
     • System Management: get_current_config, get_system_status, trigger_crawl

Any client supporting Model Context Protocol can connect to TrendRadar:

Service Address: http://localhost:3333/mcp

Basic Config Template:

{
  "name": "trendradar",
  "url": "http://localhost:3333/mcp",
  "type": "http",
  "description": "News Trending Aggregation Analysis"
}

Basic Config Template:

{
  "name": "trendradar",
  "command": "uv",
  "args": [
    "--directory",
    "/path/to/TrendRadar",
    "run",
    "python",
    "-m",
    "mcp_server.server"
  ],
  "type": "stdio"
}

Notes:

• Replace /path/to/TrendRadar with actual project path
• Windows paths use backslash escape: C:\\Users\\...
• Ensure project dependencies installed (ran setup script)

1. After registration, go to Management Dashboard at top right
2. Click API Keys on the left
3. Find default API KEY at page bottom, click eye icon to view, then copy (⚠️ Note: Don't click the copy button on the far right)

1. Open Cherry Studio, go to settings
2. Select "302.AI" as model provider
3. Paste the API Key you just copied
4. Click Manage, now you can use all supported AI models

Tip: Cherry Studio has natively integrated 302.AI, you can see the complete model list after configuration.

Q: How long does $1 free credit last? A: Depends on usage frequency and model selection, can run multiple test sessions.

Q: What after free credit runs out? A: You can top up as needed, pay-as-you-go. Major AI model prices are now relatively affordable.

Check Steps:

1. Confirm port 3333 is not occupied:

   # Windows
   netstat -ano | findstr :3333
   
   # Mac/Linux
   lsof -i :3333
   

2. Check if project dependencies installed:

   # Re-run install script
   # Windows: setup-windows.bat or setup-windows-en.bat
   # Mac/Linux: ./setup-mac.sh
   

3. View detailed error logs:

   uv run python -m mcp_server.server --transport http --port 3333
   

4. Try custom port:

   uv run python -m mcp_server.server --transport http --port 33333
   

Solutions:

1. STDIO Mode:

   • Confirm UV path correct (run which uv or where uv)
   • Confirm project path correct and no Chinese characters
   • Check client error logs

2. HTTP Mode:

   • Confirm service started (visit http://localhost:3333/mcp)
   • Check firewall settings
   • Try using 127.0.0.1 instead of localhost

3. General Checks:

   • Restart client application
   • Check MCP service logs
   • Use MCP Inspector to test connection

Possible Reasons:

1. Data Does Not Exist:

   • Confirm crawler has run (have output directory data)
   • Check query date range has data
   • Check available dates in output directory

2. Parameter Error:

   • Check date format: YYYY-MM-DD
   • Confirm correct platform ID: zhihu, weibo, etc.
   • See parameter descriptions in tool docs

3. Config Issues:

   • Confirm config/config.yaml exists
   • Confirm config/frequency_words.txt exists
   • Check config file format is correct