wechat-daily-report

WeChat group chat daily report generation Skill

Workflow

1. 运行 analyze_chat.py 分析聊天记录
   ↓
2. AI 根据聊天文本生成内容 (ai_content.json)
   ↓
3. 运行 generate_report.py 生成日报图片 (.png)

⚠️ The final output is a PNG image, not HTML. Make sure the --output parameter uses the .png suffix.

Usage steps

1. Analyze chat records

python scripts/analyze_chat.py <聊天记录.json> --output-stats stats.json --output-text simplified_chat.txt

Output:

stats.json - Statistics (chatterbox ranking, staying-up-late champion, word cloud, etc.)
simplified_chat.txt - Full chat text, for AI analysis

2. AI-generated content

Based on simplified_chat.txt and stats.json, generate AI content according to the format in references/ai_prompt.md , and save it as ai_content.json。

The content AI needs to generate includes:

topics: Discussion hot topics (3-5)
resources: Tutorials/resource sharing
important_messages: Important messages
dialogues: Interesting conversations
qas: Q&A
talker_profiles: Characteristic tags of chatterbox members (common words are already counted by the script)

3. Generate daily report image

python scripts/generate_report.py --stats stats.json --ai-content ai_content.json --output report.png --clean-temp

✅ The output suffix must be .png, using iPhone 14 Pro Max resolution (430x932 @3x)

Using the --clean-temp parameter can automatically delete temporary files after generation is complete (stats.json, ai_content.json, simplified_chat.txt, temporary HTML)

Image generation requires installing playwright:pip install playwright && playwright install chromium

Chat record JSON format

{
  "meta": {
    "name": "群名称",
    "platform": "wechat",
    "type": "group"
  },
  "members": [
    {"platformId": "xxx", "accountName": "昵称", "groupNickname": "群昵称"}
  ],
  "messages": [
    {
      "sender": "platformId",
      "accountName": "昵称",
      "groupNickname": "群昵称",
      "timestamp": 1234567890,
      "type": 0,
      "content": "消息内容"
    }
  ]
}

💡 Display name priority:groupNickname > accountName

Message type (type):

0: Plain text (the script and AI analyze this type)
1: Image
2: Voice-to-text (the script and AI analyze this type, content format:[语音转文字] xxx）
5: Animated emoji
99: System message

Script analysis content (ensures accuracy)

Data	Description
Total message count	Count of all messages
Active user count	Deduplicated user count
Time range	Time of first and last messages
Chatterbox ranking TOP3	Sorted by number of messages + common words
Staying-up-late champion	The person most active latest between 23:00-06:00
Word cloud data	jieba word segmentation + word frequency statistics

AI-generated content (requires understanding context)

Content	Input data
Discussion hot topics	Condensed text + word cloud TOP50
Member profiles	Statements in the condensed text
Interesting conversations	High-interaction segments
Q&A identification	Question mark messages ± context
Tutorials/resources	Condensed full-text identification

Dependencies

pip install jieba jinja2 playwright
playwright install chromium