MCP Neon Azure AI Agent — DevOps skill for Claude Code

Azure AI Agent Service, MCP and Neon Integration

This project connects **Azure AI Agent Service** with **Neon Serverless Postgres** using **MCP (Model Context Protocol)**. It lets you send commands to an AI agent that can create a Neon management Agent to manage Neon projects, databases, and more.


What It Does

• MCP server integrates with Azure AI Foundry to enable connections to your existing Azure AI Agents.
• MCP tool creates a new Neon management agent in the Azure AI Foundry. Starts an AI Agent with tools like create_project (via Neon API).
• Lets you ask the agent to do tasks like creating a Neon project.
• Uses MCP to communicate with the agent in real time.

Why Use This?

• Use natural language to control your Neon database
• Automate project and database management
• Build powerful RAG (Retrieval-Augmented Generation) apps using Neon

How It Works

1. When the server starts, it:
   • Loads environment variables
   • Connects to Azure AI Agent Service
   • Creates an AI Agent with tool functions (like create_project)
2. The agent listens for queries through MCP
3. When a query is sent, it:
   • Creates a thread
   • Adds a message
   • Runs the tool automatically
   • Returns the result

Requirements

• Python 3.9+
• An Azure AI Foundry setup
• A Neon API Key
• MCP installed and running. Refer to the Claude Desktop Users get started guide

Setup

1. Clone this repository:

   git clone https://github.com/neondatabase-labs/mcp-neon-azure-ai-agent.git
   cd mcp-neon-azure-ai-agent

2. Create a new virtual environment:

   python -m venv venv && source venv/bin/activate

3. Instal