How msft-innovation-hub-india/mcp-az-storage-svc-sample works
Project Type and Purpose
• This is a full-stack sample application demonstrating the Model Context Protocol (MCP) over Azure Blob Storage.
• It consists of two major deliverables:
a. An MCP Server (“mcp-server-az-storage-svc”) that exposes Azure Blob Storage operations as asynchronous Python APIs.
b. An MCP Client AI Assistant (“mcp-client-ai-assistant”) that provides an AI-powered chat interface (using Azure OpenAI GPT-4o) to discover and invoke the server’s storage tools via MCP.
File-Structure Analysis
• Top-level folders:
• mcp-server-az-storage-svc/
– Contains a Dockerfile, .dockerignore, .gitignore, plus the Python package “mcp-server” with server.py, az_storage_svcs.py, requirements.txt, and an .env_example.
• mcp-client-ai-assistant/
– Holds client.py, requirements.txt, .env_example, .gitignore and its own README.
• Patterns & configuration:
• Server packaged as a microservice (Dockerized, entry in server.py).
• Client is a standalone script/app using MCP client libraries and Azure OpenAI SDK.
• Common use of .env files for configuration, requirements.txt for dependencies.
Key Architectural Insights from READMEs
• Server:
• Implements MCP server interfaces over HTTP or WebSocket (asynchronous).
• Authenticates to Azure Blob Storage via Managed Identity or Azure CLI.
• Core module: az_storage_svcs.py (wraps list/create/delete, upload/download).
• Client:
• Discovers server-endpoints/tools at startup.
• Uses OpenAI GPT-4o to translate natural language into MCP calls.
• client.py orchestrates chat loop, calls server, formats responses.