微信消息助手 (WeChat Message Assistant)

This skill enables AI to interact with WeChat directly through AI chat, providing two core capabilities: retrieving recent messages and sending messages to contacts or groups.

Platform Requirements

• OS: Windows 10/11 only
• WeChat: WeChat PC client must be installed, running, and logged in
• Python: Python 3.11+ (automatically handled by Goodable)

Core Capabilities

1. Retrieve Recent Messages

Fetch the last 10-15 messages from any WeChat contact or group chat.

What you can ask:

Get the last 15 messages from "张三"
Show me recent messages from "Product Team" group
Retrieve messages from "李四", last 10 only

What AI will do:

• Opens the specified chat in WeChat
• Retrieves the requested number of messages (default: 15, max: 100)
• Returns messages with timestamp, sender, and content
• Formats as structured JSON for easy parsing

2. Send Messages

Send a text message to any WeChat contact or group chat.

What you can ask:

Send "Meeting at 3pm" to "张三"
Message "Product Team" group: "Release notes updated"
Tell "李四": "Please review the document"

What AI will do:

• Opens the specified chat in WeChat
• Sends the message content
• Confirms successful delivery

How to Use

Prerequisites

Before using this skill, ensure:

1. WeChat PC client is running
2. You are logged into your WeChat account
3. The contact/group name you specify exists in your WeChat

Example Usage

Scenario 1: Check messages from a colleague

User: "帮我看一下张三最近给我发了什么消息"
AI: [Runs get_messages.py script]
AI: "张三最近的5条消息：
1. [2024-01-28 10:30] 张三: 会议资料已发送
2. [2024-01-28 11:00] 张三: 请查收
..."

Scenario 2: Send a message

User: "给产品群发个消息：今天下午3点开会"
AI: [Runs send_message.py script]
AI: "消息已成功发送到产品群"

Instructions for AI

When the user requests WeChat message operations:

1. Retrieve Messages

Run the get_messages.py script:

python scripts/get_messages.py "<Contact/Group Name>" [count]

Parameters:

• Contact/Group Name: Exact name as shown in WeChat (required)
• count: Number of recent messages to retrieve (optional, default: 15, max: 100)

Output Format: The script returns JSON with message details:

[
  {
    "time": "2024-01-28 10:30:45",
    "sender": "张三",
    "content": "会议资料已发送",
    "type": "text"
  }
]

IMPORTANT - Output Formatting for User: After successfully retrieving messages, you MUST format them in a clean, readable way. DO NOT show raw JSON to the user.

For 5 or fewer messages - Use a numbered list:

📨 **张三** 的最近消息：

1. [2024-01-28 10:30] 张三: 会议资料已发送
2. [2024-01-28 11:00] 张三: 请查收
3. [2024-01-28 14:30] 你: 好的，已收到

Tips:

• Always include contact/group name in the header
• Keep timestamps concise (remove date if same day)
• Group consecutive messages from same sender when appropriate
• Highlight important information if user asks for summary

Error Handling:

• If WeChat is not running: "WeChat PC client is not running or not logged in"
• If contact not found: "Contact or group 'xxx' not found"
• Parse the error and provide user-friendly explanation in Chinese

2. Send Messages

Run the send_message.py script:

python scripts/send_message.py "<Contact/Group Name>" "<Message Content>"

Parameters:

• Contact/Group Name: Exact name as shown in WeChat (required)
• Message Content: Text message to send (required, cannot be empty)

Output:

• Success: "Message sent successfully to 'xxx'"
• Failure: Error message with details

Important Notes:

• Always enclose contact names and message content in quotes
• Contact names must match exactly as shown in WeChat
• For group chats, use the full group name
• Only text messages are supported (no images/files/emojis)

3. Best Practices

Name Matching:

• Ask user to confirm the exact contact/group name if unsure
• Suggest checking WeChat contact list for correct spelling
• Chinese names are case-sensitive and must match exactly

Error Recovery:

• If "WeChat not found" error: Ask user to start WeChat and log in
• If "Contact not found" error: Ask user to verify the name
• If script fails: Check if Python environment is properly set up

User Experience:

• Summarize retrieved messages in a readable format
• Confirm before sending messages if content seems important
• Provide context about what happened after each operation

Technical Details

Script Locations

skills/goodqunbot/
├── SKILL.md                    # This file
├── scripts/
│   ├── get_messages.py         # Retrieve messages
│   └── send_message.py         # Send messages
└── wxauto_lib/                 # WeChat automation library (pyc bytecode)

Implementation

Both scripts use the wxauto library (RPA-based WeChat automation):

get_messages.py:

• Initializes WeChat instance
• Opens specified chat with wx.ChatWith(who)
• Retrieves all messages with wx.GetAllMessage(savepic=False, savefile=False)
• Returns recent N messages as JSON

send_message.py:

• Initializes WeChat instance
• Sends message with wx.SendMsg(msg=message, who=who, clear=True)
• Returns success/failure status

Limitations

1. Platform: Windows only (wxauto uses Windows API)
2. Message Types: Text only (no images, files, or rich media)
3. Rate Limits: May be subject to WeChat's anti-spam mechanisms
4. Accuracy: Contact name must match exactly (case-sensitive)
5. State Dependency: Requires WeChat PC client to be running and logged in

Troubleshooting

Privacy & Security

• Local Only: All operations are local, no data sent to external servers
• Read-Only Access: get_messages.py only reads, doesn't modify chat history
• User Control: send_message.py only sends when explicitly requested
• No Logging: Messages are not stored or logged by the scripts
• Compliance: Use responsibly and comply with WeChat Terms of Service

Related Features

This skill complements the GoodQunBot App (template.json), which provides:

• Web UI for browsing all groups
• AI-powered chat summaries
• Data visualization (activity charts, word clouds)
• Member management and business opportunity discovery

Skill vs App:

• Skill Mode: Quick AI-driven message operations in chat
• App Mode: Full-featured web interface for comprehensive chat management
• Both modes share the same wxauto library and work independently