TikTok Automation Skill 🎵

Production-ready automation untuk TikTok content posting & management.

🎯 Features

• Auto-login dengan session persistence
• Content upload (video, caption, hashtags)
• Dynamic element detection - adaptif terhadap class name changes
• Selector learning - save successful selectors ke selectors.json
• Fallback mechanism - retry dengan alternative selectors
• Progress tracking - real-time status updates

📦 Files

tiktok-automation/
├── SKILL.md              # This file
├── script.sh             # Main automation (cross-platform)
├── script.ps1            # Windows PowerShell version
├── selectors.json        # Learned element selectors
├── config.json           # Configuration (credentials, settings)
└── assets/
    └── sample-video.mp4  # Test video

🔧 Setup

1. Configure Credentials

Edit config.json:

{
  "credentials": {
    "username": "your-tiktok-username",
    "password": "your-password",
    "sessionFile": "tiktok-session.json"
  },
  "upload": {
    "defaultPrivacy": "public",
    "allowComments": true,
    "allowDuet": true,
    "allowStitch": true
  },
  "browser": {
    "headless": false,
    "timeout": 30000
  }
}

2. Run the Skill

# Basic upload (Linux/macOS)
./script.sh --video path/to/video.mp4 --caption "My video"

# With hashtags
./script.sh --video path/to/video.mp4 --caption "Check this" --tags "#viral #fyp"

# Update selectors (re-learning mode)
./script.sh --update-selectors

Windows (PowerShell):

.\script.ps1 -Video "path\to\video.mp4" -Caption "My video"

🔄 How It Works

Flow:

1. Load session - Check for existing TikTok session
2. Login if needed - Auto-login with credentials
3. Navigate to upload - Go to upload page
4. Dynamic detection - Find upload button using multiple strategies:
   • Try learned selectors from selectors.json
   • Fallback to text/role-based detection
   • Fallback to structure-based detection
5. Upload video - Drag & drop or file input
6. Add metadata - Caption, hashtags, settings
7. Publish - Submit and confirm
8. Save selectors - Update selectors.json with successful selectors

Selector Learning:

Every successful element interaction is saved:

{
  "uploadButton": {
    "selector": "[data-e2e='upload-btn']",
    "method": "dataAttribute",
    "confidence": 1.0,
    "lastUsed": "2026-02-18T22:00:00Z",
    "fallbacks": [
      {"selector": "button[type='file']", "method": "css"},
      {"text": "Upload", "method": "text"}
    ]
  }
}

🛠️ Troubleshooting

TikTok Changed Their UI Again?

Run with --update-selectors flag to re-learn element locations.

Login Failed?

• Check credentials in config.json
• Clear session file: delete tiktok-session.json
• Try manual login first, then run skill

Upload Stuck?

• Check video format (MP4, MOV, WebM)
• Ensure file size < 287.6 MB
• Verify internet connection

📊 Logs

All actions logged to:

• Console output (real-time)
• tiktok-automation.log (persistent)

⚠️ Warnings

• Rate limiting: Don't upload too frequently (max 10 videos/hour)
• Terms of Service: Use responsibly, follow TikTok's ToS
• Account safety: Consider using a burner account for automation

🚀 Next Steps

Planned improvements:

• Schedule posts for optimal timing
• Analytics tracking (views, likes, shares)
• Bulk upload from folder
• Auto-generate captions with AI
• Cross-post to Instagram Reels & YouTube Shorts

---

Berkah Karya ⚡ | Part of 1ai-skills collection