wechat-desktop-sender（微信桌面发送器）

wechat-desktop-sender

Use this skill for Windows 微信桌面端发消息 workflows.

What this skill provides

Bundled scripts:

• scripts/wechat_send.py — formal single-send command entry
• scripts/wechat_send_hello.py — core implementation (open WeChat, search chat, send, verify)
• scripts/wechat_send_batch.py — serial batch sender for multiple contacts/groups
• scripts/wechat_send_campaign.py — personalized serial sender (different message per contact)
• scripts/wechat_send_template_campaign.py — template-variable sender (same template, different row values)

Quick start

Install dependencies:

pip install pywinauto pyperclip psutil pywin32 pillow pytesseract

pytesseract is optional. Install it only if OCR verification is needed.

Single send

python scripts/wechat_send.py --to "文件传输助手" --message "你好"

Useful flags:

python scripts/wechat_send.py --to "李义" --message "我是龙虾Koi，打个招呼。" --verify-title
python scripts/wechat_send.py --to "文件传输助手" --message "OCR测试" --ocr-verify

Serial batch send

python scripts/wechat_send_batch.py --to "张三,李四,文件传输助手" --message "晚上 8 点开会"

Or from file:

python scripts/wechat_send_batch.py --contacts-file contacts.txt --message "晚上 8 点开会"

Personalized send

python scripts/wechat_send_campaign.py --csv contacts_messages.csv --verify-title
python scripts/wechat_send_campaign.py --json contacts_messages.json --verify-title

Template-variable send

python scripts/wechat_send_template_campaign.py --csv template_contacts.csv --template "{name}你好，我是Koi，关于{company}这边和你打个招呼。" --verify-title

Recommended workflow

1. Ensure WeChat desktop is logged in
2. Test on 文件传输助手 first
3. Use single-send for one contact/group
4. Use batch-send for serial outreach
5. Check wechat_automation_logs/ for logs and summaries

Verification model

Current verification layers:

1. UI element location succeeded
2. Message was sent through the input box
3. Window text verification checks whether the message text appears in current WeChat window
4. Optional OCR verification if --ocr-verify is enabled and OCR deps are installed

Treat this as practical desktop automation, not a cryptographic delivery guarantee.

Boundaries

Do not claim support for these unless separately extended:

• Moments / 朋友圈 scraping
• Reliable historical message forwarding
• Multi-window true parallel sending
• Full desktop social graph extraction

Files and outputs

By default, scripts create logs under:

wechat_automation_logs/

Outputs may include:

• run logs
• failure screenshots
• control-tree dumps
• batch summary JSON files

OpenClaw workflow usage

Once this skill is installed, the user can express intent conversationally and the agent should route to the correct script:

• “给张三发一句话：今晚 8 点开会” → use single-send
• “给这 20 个人发同一条消息” → use batch-send
• “按这份名单分别发不同文案” → use personalized-send
• “按这份名单套一个模板变量群发” → use template-variable send

Preferred operational pattern:

1. Normalize the recipient list and message payload
2. If one recipient and one message → scripts/wechat_send.py
3. If many recipients and same message → scripts/wechat_send_batch.py
4. If many recipients and different messages → scripts/wechat_send_campaign.py
5. If many recipients share one template with per-row fields → scripts/wechat_send_template_campaign.py
6. Report success count, verification count, retry recommendations, and log/summary paths back to the user

When to read more

For detailed CLI usage and arguments, read:

• references/single-send.md
• references/batch-send.md
• references/personalized-send.md
• references/template-send.md