Sales Analysis MCP Server

For a comprehensive view of data

See ZAVA RETAIL DATABASE DOCUMENTATION

A Model Context Protocol (MCP) server that provides comprehensive customer sales database access for Zava Retail DIY Business. This server enables AI assistants to query and analyze retail sales data through a secure, schema-aware interface.

Prerequisites

1. Docker Desktop installed
2. Git installed
3. Azure CLI: Install and authenticate with Azure CLI
4. Access to OpenAI text-embedding-3-small model and optionally gpt-4o-mini model.

Getting Started

Open a terminal window and running the following commands:

1. Authenticate with Azure CLI

   az login
   

2. Clone the repository

   git clone https://github.com/gloveboxes/Zava-MCP-Server-and-PostgreSQL-Sample
   

3. Navigate to the project directory

   cd Zava-MCP-Server-and-PostgreSQL-Sample
   

Deploy Azure Resources

Run the following scripts to automate the deployment of Azure resources needed for the MCP server.

The deployment scripts will automatically deploy:

• Azure AI Foundry with both text-embedding-3-small and gpt-4o-mini models
• Azure Database for PostgreSQL (Flexible Server) with pgvector extension
• Application Insights for monitoring
• Service principal with appropriate permissions

The deployment will automatically initialize the PostgreSQL database with the Zava retail sample data.

Choose the script for your platform:

Windows (PowerShell)

# Run from the project root directory
cd infra && ./deploy.ps1

macOS/Linux (Bash)

# Run from the project root directory
cd infra && ./deploy.sh

Running the MCP Server

After deploying Azure resources using the deployment scripts above, the MCP server will automatically connect to your Azure PostgreSQL database using the configuration in .env.

To run the MCP servers:

# Start the MCP servers (they will connect to Azure PostgreSQL)
# Option 1: Using VS Code tasks
# Use Ctrl+Shift+P -> "Tasks: Run Task" -> "Start Supplier and Finance MCP Servers"

# Option 2: Using terminal
python -m app.mcp.supplier_server &
python -m app.mcp.finance_server &

The deployment script automatically creates a .env file with the correct Azure connection details.

Manual Database Initialization

If the automatic database initialization fails during deployment, you can initialize it manually:

cd infra
./init-azure-db.sh

This script will:

• Test the connection to Azure PostgreSQL
• Enable required extensions (pgvector, plpgsql)
• Create the store_manager user
• Restore the Zava retail database from the backup file
• Set up proper permissions

Cleanup Azure Resources

When you’re done with the workshop, you can remove all Azure resources:

cd infra
./cleanup.sh