Explosive Content Detector for WeChat Official Accounts

The Core Problem

A primary pain point for content creators on WeChat is the difficulty in uncovering trending topics and identifying potential "hit" articles manually. In the vast WeChat ecosystem, sifting through content to find pieces with high engagement potential is time-consuming, causing creators and operators to miss critical windows. The fundamental need is to answer: "Which sub-topic is gaining traction?" and "What makes a competitor's article successful?" This demands a tool that goes beyond simple search to quantify content value.

How the Skill Works

This skill provides an intelligent query solution powered by the Redfox API. The core workflow is:

1. Intelligent Retrieval and Structured Ranking

   • Accepts one or more keywords (comma-separated) via the --keyword parameter to retrieve matching articles.
   • Instead of a flat list, results are ranked using a Three-Dimensional Scoring Model:
     • Relevance (Max 10 points): Measures how closely an article matches the query.
     • Hotness (Max 3 points): Reflects the article's immediate engagement and spread metrics.
     • Timeliness (Max 2 points): Weights newer content more heavily.
   • A total score (Max 15 points) is calculated, and articles are presented in descending order. For articles with over 5,000 reads, the count is auto-formatted (e.g., 10.0w) for readability.

2. Flexible Contextual Expansion

   • Automatic Time Expansion: When data from the default 7-day window is insufficient, the system automatically expands the time range in sequence: 1 day -> 7 days -> 30 days. This ensures data is retrieved without manual user intervention.
   • Sector Keyword Identification & Expansion: For broad sector keywords like "career" or "relationships", the skill identifies them and proactively suggests 10 sub-directions (e.g., "managing superiors," "workplace attire"), enabling precise, drill-down searches. This expansion requires explicit user confirmation.
   • Cold-Start Recommendations: When a search yields zero or few results, the system automatically provides a list of recent popular articles and hot topics as alternatives, preventing dead ends.

3. Data Output and Subscription

   • Core results are returned as structured JSON via standard output (stdout), ideal for programmatic parsing.
   • An optional HTML report file can be generated for standalone viewing and sharing.
   • After a query, a calendar subscription task (via the calendar_create tool) can be set up to automatically push the latest hit article list at a scheduled time.

Use Cases & Boundaries

• Ideal For: Discovering topics within a niche, analyzing a competitor's content strategy, quickly scanning trends across multiple sectors, and establishing a daily hotspot tracking mechanism.
• Not Suitable For: Querying articles older than 30 days, obtaining real-time (post-ingestion) engagement data, or analyzing "long-tail" content with fewer than 5,000 reads.
• Key Limitations: Data updates daily at 7 AM, so querying "today" yields yesterday's data. All scoring and ranking are based on historical ingested data and do not reflect an article's current, real-time status.