Snowflake Multi-Tool Agent Framework

A powerful and flexible orchestration framework that combines multiple Snowflake-powered tools into a unified agent system, enabling natural language interactions with various data sources and services.

🌟 Features

• Interactive Streamlit Interface: Modern web interface for real-time interaction with the agent system
• Multiple Specialized Tools:
  • 📊 Cortex Analyst Tool: Analyzes sales metrics and performance data
  • 🔍 Knowledge Base Search Tool: Searches through internal documentation and articles
  • 📈 Stock Market Tool: Retrieves real-time stock market data using yfinance
• Asynchronous Processing: Handles multiple requests efficiently with async operations
• Real-time Logging: Live feedback on tool selection and processing status
• Chat History Management: Save and download conversation history
• Snowflake Integration: Seamless connection with Snowflake data warehouse

🚀 Getting Started

Prerequisites

• Python 3.11.10
• Snowflake account with appropriate access
• Required Python packages (install via pip):

  orchestration-framework@git+https://github.com/Snowflake-Labs/orchestration-framework.git
  streamlit
  snowflake-ml-python
  python-dotenv
  yfinance
  requests
  

Configuration

1. Clone this repository
2. Copy .env_example to .env and fill in your Snowflake credentials:

   SNOWFLAKE_ROLE="<ROLE>"
   SNOWFLAKE_WAREHOUSE="<WAREHOUSE>"
   SNOWFLAKE_USER="<USER>"
   SNOWFLAKE_PASSWORD="<PASSWORD>"  # Remove if using SSO
   SNOWFLAKE_AUTHENTICATOR="externalbrowser"  # For SSO
   SNOWFLAKE_ACCOUNT="<ACCOUNT>"
   SNOWFLAKE_DATABASE="<DATABASE>"
   SNOWFLAKE_SCHEMA="<SCHEMA>"

Running the Application

1. Start the Streamlit app:

   streamlit run streamlit.py
   

2. Access the web interface through your browser (typically http://localhost:8501)

💡 Usage

The framework provides a chat-like interface where you can:

1. Query Sales Data: Ask questions about sales metrics, revenue, marketing spend, etc.
2. Search Knowledge Base: Look up information in internal documentation
3. Check Stock Data: Get real-time stock market information by ticker symbol

Example queries:

• "What is the total revenue for the toy category?"
• "Show me marketing campaign details for the beauty products"
• "Get the current stock price for SNOW"

🛠 Architecture

The framework consists of three main components:

1. Agent Gateway: Core orchestration layer that routes requests to appropriate tools
2. Specialized Tools:
   • CortexAnalystTool: Handles structured data analysis
   • CortexSearchTool: Manages unstructured data searches
   • PythonTool: Executes custom Python functions (e.g., stock data retrieval)
3. Streamlit Interface: Web-based user interface with real-time updates

📝 Development

The project includes:

• streamlit.py: Main application file with UI and tool integration
• Notebook.ipynb: Jupyter notebook for development and testing
• Configuration files (.env_example, .gitignore, etc.)

🔒 Security

• Sensitive credentials are managed through environment variables
• SSO authentication support for Snowflake
• User input validation and sanitization

📄 License

Copyright 2024 Snowflake Inc. Licensed under the Apache License, Version 2.0