Topic Planning + Content Calendar

Use Cases

• User says "帮我规划下周公众号内容"
• User says "最近有什么热门选题可以写"
• User says "帮我做一份内容日历"
• User wants to know what competitor accounts are writing about
• Need to make topic decisions based on data

Dependencies

Node.js + cheerio (install once):

npm install -g cheerio

Script Directory

Script	Purpose	Usage
scripts/wechat_search.js	Sogou WeChat article search	node "$SKILLS_ROOT/content-planner/scripts/wechat_search.js" "keyword"

Search Script Parameters

IMPORTANT: Always use the $SKILLS_ROOT environment variable to locate scripts.

# Basic search
node "$SKILLS_ROOT/content-planner/scripts/wechat_search.js" "keyword"

# Limit result count
node "$SKILLS_ROOT/content-planner/scripts/wechat_search.js" "keyword" -n 15

# Save to file
node "$SKILLS_ROOT/content-planner/scripts/wechat_search.js" "keyword" -n 20 -o result.json

# Parse real URLs (extra network requests, may be blocked by anti-scraping)
node "$SKILLS_ROOT/content-planner/scripts/wechat_search.js" "keyword" -n 5 -r

Output Fields: Article title, article URL, article summary, publish time, source account name

Workflow

Step 1: Clarify Planning Scope

Confirm the following information with the user (ask all at once):

帮你规划内容，先确认几件事：

1. 规划周期？（本周 / 下周 / 自定义时间范围）
2. 有没有特定想写的方向或关键词？
3. 每周几篇？（默认3篇）

Step 2: Trending Scan

Execute multiple rounds of searches covering different dimensions:

Search Strategy:

1. Core domain keyword search — Search with 2-3 core keywords related to the account's field
2. User-specified keyword search — If user has specific directions
3. General trending search — Search with combinations of "热点", "热门", "最新" with domain keywords

# Example: Tech domain
node "$SKILLS_ROOT/content-planner/scripts/wechat_search.js" "AI 最新趋势" -n 10
node "$SKILLS_ROOT/content-planner/scripts/wechat_search.js" "大模型应用" -n 10
node "$SKILLS_ROOT/content-planner/scripts/wechat_search.js" "科技热点 2026" -n 10

Wait 3-5 seconds between each search to avoid triggering anti-scraping mechanisms.

Step 3: Competitor Analysis

Extract from search results:

Analysis Dimension	Extracted Content
Title Strategy	Which title patterns get high engagement
Topic Direction	Which directions are recent hot topics
Content Angle	What angles do existing articles take, how to differentiate
Publish Time	Competitors' publishing frequency and timing

Step 4: Generate Topic Recommendations

Based on trending data and competitor analysis, generate 5-10 topic suggestions. Each topic must include:

• Alternative Titles (2 styles)
• Target Audience
• Content Angle (differentiation point)
• Recommended Style: deep-analysis / practical-guide / story-driven / opinion / news-brief
• Urgency: 🔥 Urgent / 📅 This week / 📦 Reserve
• Reference Articles (from search results)

Topic Quality Requirements:

• Each topic must be based on real search data, not fabricated
• Each topic must have a clear differentiation angle
• Must include at least 1 high-urgency topic (🔥) and 2 reserve topics (📦)

Step 5: User Approval

This is a mandatory approval gate and cannot be skipped.

Present the topic list and ask user to select, adjust, and schedule.

Step 6: Generate Content Calendar

Output content_calendar.json:

{
  "week": "2026-W13",
  "created_at": "2026-03-25",
  "articles": [
    {
      "id": 1,
      "date": "2026-03-26",
      "day": "Wednesday",
      "topic": "Topic Title",
      "angle": "Differentiation Angle",
      "style": "deep-analysis",
      "audience": "Target Audience",
      "urgency": "this-week",
      "status": "planned",
      "keywords": ["keyword1", "keyword2"]
    }
  ]
}

Step 7: Output Confirmation

Present a summary table and prompt user to start writing with article-writer skill.

Search Considerations

• Search results may be empty (anti-scraping), retry with different keywords
• Multiple searches in short time may trigger restrictions, recommend 3-5 second intervals
• The -r parameter for parsing real URLs has low success rate, avoid unless necessary

Integration with Other Skills

• The generated content_calendar.json is the input source for article-writer
• article-writer reads entries with status planned from the calendar