Export your best conversation pairs from OpenWebUI to AI Curator for LLM fine-tuning dataset management.

• One-Click Export: Capture high-quality conversations with a single button click
• Smart Metadata: Auto-detects model info and file attachments
• Pre-configured Settings: Set default category, quality rating, and dataset in the plugin settings
• Duplicate Detection: Automatically skips duplicate conversations
• Privacy First: Your data goes directly to your local EdukaAI instance
• Docker Compatible: Works seamlessly with Docker-based OpenWebUI installations

• AI Curator running locally (default: http://localhost:3333)
• OpenWebUI >= 0.3.0

1. Install the Plugin

   • Go to OpenWebUI → Admin Panel → Functions
   • Click "New Function" → Import from Link
   • Paste this URL:

     https://github.com/ElGap/edukaai-openwebui
     

2. Configure AI Curator Connection

   • Click the gear icon next to the AI Curator Export function
   • Set your AI Curator endpoint:
     • Docker + macOS/Windows: http://host.docker.internal:3333
     • Docker + Linux: Use your host IP (e.g., http://172.17.0.1:3333)
     • Native install: http://localhost:3333
   • Configure default category, quality rating, and dataset ID
   • Save

3. Enable the Plugin

   • Toggle the AI Curator Export function to "On"
   • Click on More and make sure Global is also set to "On"

1. Have a conversation in OpenWebUI
2. When you see a good exchange you want to save, look for the 📚 Export to AI Curator button under the assistant's message
3. Click the button - it will immediately export using your pre-configured settings
4. Get instant feedback: "✅ Exported to AI Curator!"
5. Open AI Curator at http://localhost:3333 to view and manage your saved samples

Tip: The export uses your default settings (category, quality, dataset). You can change these defaults in the plugin configuration.

Configure via the Valves (settings) in OpenWebUI:

Important: The EDUKAAI_ENDPOINT must match your setup:

• Docker on macOS/Windows: http://host.docker.internal:3333 (default, should work)
• Docker on Linux: Use http://YOUR_HOST_IP:3333 (find IP with ip addr show docker0)
• Native installation: http://localhost:3333

The plugin captures:

• User's question → instruction field
• Assistant's response → output field
• System prompt (if any) → systemPrompt field
• Model name → context.model.name
• File attachments → context.files array

Data is sent to EdukaAI's /api/capture endpoint via REST API.

Exported records follow the EdukaAI Universal Record format:

{
  "source": "openwebui",
  "apiVersion": "1.0",
  "records": [{
    "instruction": "User's question here",
    "output": "Assistant's response here",
    "systemPrompt": "Optional system instructions",
    "category": "coding",
    "qualityRating": 4,
    "tags": ["python", "debugging"],
    "context": {
      "model": {"name": "gpt-4"},
      "files": []
    }
  }],
  "options": {
    "autoApprove": false,
    "skipDuplicates": true
  }
}

1. Clone this repository:

   git clone https://github.com/ElGap/edukaai-openwebui.git
   cd edukaai-openwebui

2. Copy the function to your OpenWebUI instance for testing

3. Make changes and test locally

edukaai-openwebui/
├── edukaai_openwebui.py    # Main plugin file
├── README.md               # This file
├── LICENSE                 # MIT License
└── .github/                # GitHub templates and workflows

Contributions are welcome! Please feel free to submit a Pull Request.

1. Fork the repository
2. Create your feature branch (git checkout -b feature/AmazingFeature)
3. Commit your changes (git commit -m 'Add some AmazingFeature')
4. Push to the branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

This project is licensed under the MIT License - see the LICENSE file for details.

• AI Curator Repository
• OpenWebUI

• Built for the AI Curator and EdukaAI project
• Inspired by the need for simple, privacy-focused LLM training data collection
• Thanks to the OpenWebUI team for the excellent plugin system