Example

Example Implementation

Under the app folder, you'll find a simple Streamlit application that demonstrates how to build a compatible backend:

Features

• Uses llama-index for document embedding and retrieval
• Integrates with Google Gemini for LLM responses
• Loads documentation from your docs/ directory
• Provides a chat interface optimized for iframe embedding

Setup Instructions

1. Install dependencies:

   cd app
   uv sync  # or pip install -r requirements.txt
   

2. Configure API keys:

3. The example uses Google Gemini API
4. Set your API key in app/main.py (or use environment variables for production)

5. You'll need:

   • Google Gemini API key for the LLM
   • Google Gemini API key for embeddings

6. Configure the documentation path:

7. Update DATA_PATH in app/main.py to point to your documentation directory

8. By default, it points to ../docs/ relative to the app directory

9. Run the Streamlit app:

   streamlit run app/main.py
   

   The app will typically run on http://localhost:8501

10. Configure MkDocs:

    plugins:
        - chatbot:
            url: "http://localhost:8501"  # or your production URL
    

Building Your Own Backend

You're not limited to the Streamlit example. You can build your backend using any framework (Flask, FastAPI, Django, etc.) as long as it:

• Serves a web page that can be embedded in an iframe
• Provides a chat interface (can be simple HTML/JavaScript or a full framework)
• Connects to an embedding service and LLM API
• Processes your documentation files for semantic search

The backend can be: - Hosted separately - Deploy your chat backend independently (e.g., on a cloud service) - Co-located with MkDocs - Run both services on the same server - Multi-tenant - Support multiple documentation projects using the project query parameter