[Skill Description] An intelligent assistant that automatically analyzes Weibo trending topics and transforms them into product ideas. By scraping the Weibo trending list in real time, it deeply mines the background information and event context of each hot topic, uses AI to extract product ideas from the perspective of "80% fun + 20% useful," and ultimately generates a visualized HTML analysis report. It helps product managers, entrepreneurs, and developers quickly discover market opportunities and transform hot topics into executable product plans.
[Core Capabilities] - Real-time trending scraping: Automatically obtains the Weibo trending TOP50 list through the Tianju Shuxing API, including trending terms, popularity indexes, and tags - Intelligent background search: Uses the Web Search tool to search for related news, event context, and social background for each trending topic - AI idea analysis: Based on hot topic information, extracts product ideas using a multi-dimensional scoring system of "80 points for fun + 20 points for usefulness" - Multi-idea generation: Generates multiple product idea plans for each hot topic, including name, core features, and target user description - Visualized report: Generates a responsive HTML report with a vibrant orange theme, supports viewing on mobile phones/tablets, and highlights high-scoring ideas - Intelligent scoring and grading: Automatically calculates comprehensive scores, marks >80 points as excellent, and 60-80 points as good
[Execution Flow]
Step 1: Scrape the Weibo trending list
- Run python scripts/fetch_weibo_hot.py Call the Tianju Shuxing API
- Obtain TOP50 trending data: trending terms, popularity indexes, and trending tags
- Parse the JSON-formatted returned results and verify data integrity
Step 2: Deeply search hot topic backgrounds
- Use the Web Search tool to search for related news for each trending term
- Search keyword format: "<trending term> event background latest developments"
- Collect key information from the search results: event timeline, involved entities, social impact, and hot topics of user discussion
Step 3: AI analysis to generate product ideas
- Based on hot topic background information, analyze productization opportunity points
- Generate multiple product ideas from the following dimensions (at least 3-5):
* **Fun analysis** (80-point weight): idea novelty, topic popularity, user participation, and dissemination potential
* **Usefulness analysis** (20-point weight): practical value, strength of user demand, and degree of market pain point resolution
- Generate structured information for each idea:
* Product name (concise and memorable, reflecting hot topic elements)
* Core features (3-5 key feature points)
* Target user profile (age, interests, demand scenarios)
* Comprehensive score (fun × 0.8 + usefulness × 0.2)
Step 4: Generate the HTML analysis report
- Read the [templates/report_template.html](templates/report_template.html) template
- Fill the following content into the template:
1. **Report header**: generation date, total number of hot topics, analysis time
2. **Hot topic overview**: quick preview of the TOP10 list
3. **Detailed analysis**: complete information for each hot topic
- Trending term, popularity index, tags
- Event context (in timeline form)
- Product idea list (sorted from high to low score)
- Score grade labels (excellent >80 points, good 60-80 points, average <60 points)
4. **Creative Ranking**: All creatives are displayed in descending order of score
5. **Data Statistics**: Number of high-scoring creatives, average score, popularity distribution
- Apply responsive styles and vibrant orange color scheme
- **Automatically create directory structure** and save the file:
```bash
BASE_DIR="/Users/wanglingwei/Movies/violinvault/SynologyDrive/Clipping/19-ClaudeCode/微博热搜"
# 创建 YYYY/MM/ 目录结构
OUTPUT_DIR="${BASE_DIR}/$(date +%Y/%m)"
mkdir -p "${OUTPUT_DIR}"
# 输出文件
OUTPUT_FILE="${OUTPUT_DIR}/$(date +%Y-%m-%d)_weibo_hotspot_report.html"
```
- Output file name format:`<YYYY-MM-DD>_weibo_hotspot_report.html`
- Full path example:`/Users/wanglingwei/Movies/violinvault/SynologyDrive/Clipping/19-ClaudeCode/微博热搜/2026/01/2026-01-04_weibo_hotspot_report.html`
Step 5: Complete and output
- Confirm that the HTML file was generated successfully
- Confirm that the directory structure was created correctly (year/month folders)
- Return: "✅ Weibo hot search product creative report has been generated"
- Provide the full file path (including the year-month directory)
- Prompt the user to open the corresponding directory directly in Finder to view it
[Notes]
- Output directory structure: Reports are automatically organized by year and month into /Users/wanglingwei/Movies/violinvault/SynologyDrive/Clipping/19-ClaudeCode/微博热搜/YYYY/MM/ directory, with one folder per month, making management and archiving easy
- Automatic directory creation: Use mkdir -p command to automatically create year-month directories that do not exist, with no need to create them manually
- File overwrite notice: Running multiple times on the same day will overwrite the previous report; if you need to keep it, please manually rename the file
- API call limit: Tianju Shuxing API has a daily call limit (regular members: 100 times/day), so control the analysis frequency reasonably
- Search depth control: It is recommended to search 3-5 related news items for each hot search to avoid excessive searching consuming tokens
- Scoring objectivity: AI scoring is based on information completeness; if background information is insufficient, scoring bias may occur and can be adjusted manually
- Creative diversity: Ensure differentiated creatives are generated for different types of hot searches to avoid template-based output
- HTML file encoding: Save using UTF-8 encoding to ensure Chinese displays correctly
- Report update strategy: It is recommended to analyze once per day to avoid repeatedly analyzing the same hot search
- Vercel deployment: The generated HTML can be dragged directly to Vercel for deployment without additional configuration
- Data timeliness: Weibo hot searches update every 30 minutes, so the report generation timestamp is important
- Sensitive content filtering: If an illegal or non-compliant hot search is encountered, skip the analysis and log it
- Scoring system explanation: Interest level has a higher weight (80%), because trending products usually rely on topicality and spreadability