A Node.js-based Model Context Protocol (MCP) server that connects your warehouse operations in Microsoft Dynamics 365 Supply Chain Management (D365 SCM) to AI agents.

This server runs on Azure Web App and exposes warehouse operations as AI-callable tools:

• readOpenLines — Get open outbound pick orders
• confirmWork — Confirm completed warehouse work

---

🚀 Overview

This project acts as a bridge between:

• AI agents in Azure AI Foundry or Copilot Studio
• D365 SCM warehouse APIs (MHAX)
• MCP tool execution layer on Azure Web App It enables AI-driven warehouse operations like:
• Reading pick work queues from D365
• Confirming warehouse work execution via AI
• Automating logistics decisions in real-time

---

🏗️ Architecture

Warehouse User (Teams / App)
        ↓
Azure AI Foundry Agent or Copilot Studio Agent
        ↓
MCP Server (Azure Web App)
        ↓
D365 SCM (MHAX / REST APIs)

---

📦 Features

✅ MCP Streamable HTTP server — Compatible with Copilot Studio & Azure AI Foundry
✅ OAuth 2.0 client credentials — Secure D365 authentication
✅ D365 SCM integration — Direct access to warehouse APIs
✅ Azure App Service ready — Deploy in minutes
✅ Session-based execution — Stateful MCP sessions
✅ REST fallback endpoints — Test with Postman or REST Client
✅ Secure configuration — Environment-based secrets

---

🧰 Prerequisites

Required tools

• Node.js 22+
• Git
• Azure CLI — for deployment
• VS Code with Azure App Service extension

Azure & Copilot Studio requirements

• Azure subscription or Microsoft 365 with Copilot Studio
• Permission to create:
  • App Registrations (Azure AD)
  • App Services (Azure Web App)
• Access to D365 SCM environment with admin rights

---

📁 Project Structure

src/
├── server.js              (Main MCP server)
├── tools/
│   ├── registry.js        (Tool catalog)
│   ├── readLines.js       (Tool: readOpenLines)
│   └── confirmWork.js     (Tool: confirmWork)
├── auth/
│   └── d365Auth.js        (OAuth 2.0 token management)
└── middleware/
    └── validate.js        (Request validation)
 
.env                       (Your secrets - DO NOT COMMIT)
.env.example              (Template)
package.json
test.http                 (REST Client test file)

---

⚙️ Installation

git clone https://github.com/<your-org>/<repo>.git
cd <repo>
npm install

---

🔐 Environment Configuration

Create .env file

cp .env.example .env

Required variables

D365_TENANT_ID=xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx
D365_CLIENT_ID=xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx
D365_CLIENT_SECRET=your-secret-value-here
D365_BASE_URL=https://yourcompany.operations.dynamics.com
PORT=3000
SERVER_API_KEY=your-generated-key-optional

Variable reference

Variable	Description	Where to get it
D365_TENANT_ID	Azure AD tenant ID	Azure Portal → Azure Active Directory → Overview
D365_CLIENT_ID	App registration client ID	Azure Portal → App Registrations → Your app
D365_CLIENT_SECRET	OAuth secret	Azure Portal → App Registrations → Certificates & secrets
D365_BASE_URL	D365 environment URL	Your D365 domain (no trailing slash)
SERVER_API_KEY	Optional security key	Generate: node -e "console.log(require('crypto').randomBytes(32).toString('hex'))"

---

🔑 Azure App Registration Setup

Create an app in Azure Portal:

Steps:

1. Go to Azure Portal → Azure Active Directory → App registrations → + New registration
2. Fill in details and register
3. Copy:
   • Tenant ID (from Overview)
   • Client ID (from Overview)
4. Go to Certificates & secrets → + New client secret → copy Value
5. Go to API permissions → + Add a permission
   • Search: Dynamics ERP
   • Select: Microsoft Dynamics ERP
   • Check: AX.FullAccess
   • Click: Grant admin consent ✓

---

🏢 D365 Configuration

In Microsoft Dynamics 365:

1. Log in as admin
2. Go to: Azure Active Directory applications
3. Click: + New
4. Paste your Client ID
5. Set User ID to a service account
6. Click: Save

---

▶️ Run Locally

npm start

Expected output:

✅ Startup: D365 OAuth credentials verified.
🚀 MCP Server v3.1.0 listening on port 3000

Verify it works:

curl http://localhost:3000/health

Response:

{
  "status": "healthy",
  "server": "RJ D365 MHAX MCP Server",
  "version": "3.1.0"
}

---

☁️ Deploy to Azure Web App

Option 1: From VS Code (Easiest)

1. Install Azure App Service extension
2. Sign in to Azure
3. Press Ctrl+Shift+P → "Azure App Service: Deploy to Web App"
4. Select folder → subscription → Web App

Option 2: Azure CLI

# Create Web App
az webapp create \
  --name your-mcp-app-name \
  --resource-group your-resource-group \
  --plan your-app-service-plan \
  --runtime "NODE:22-lts"
 
# Deploy
zip -r deploy.zip . -x "node_modules/*" ".env" ".git/*"
az webapp deployment source config-zip \
  --resource-group your-resource-group \
  --name your-mcp-app-name \
  --src deploy.zip

---

⚙️ Azure Web App Configuration

In Azure Portal → Your Web App:

Application Settings

Add these environment variables:

• D365_TENANT_ID = your value
• D365_CLIENT_ID = your value
• D365_CLIENT_SECRET = your value
• D365_BASE_URL = your value
• WEBSITE_NODE_DEFAULT_VERSION = ~22

General Settings

• Startup Command: node src/server.js
• Always On: Enable (required for MCP sessions)
• Node Version: 22 LTS

---

🔗 MCP Endpoint

After deployment, your MCP server is available at:

https://your-mcp-app-name.azurewebsites.net/mcp

Use this URL when connecting to:

• Azure AI Foundry Agents
• Copilot Studio

---

🔌 Connect to Agents

Azure AI Foundry

1. Go to: Build → Agents → Your agent
2. Tools → Add MCP tool
3. Enter URL: https://your-app.azurewebsites.net/mcp
4. Click: Create
5. Ref:https://learn.microsoft.com/en-us/azure/foundry/agents/how-to/tools/model-context-protocol?pivots=java

Copilot Studio

1. Go to: Tools → Add a tool → New tool → Model Context Protocol
2. Fill in:
   • Name: RJ D365 MCP Server
   • Description: Reads pick orders and confirms warehouse work from D365
   • URL: https://your-app.azurewebsites.net/mcp
3. Click: Create
4. Ref: https://learn.microsoft.com/en-us/microsoft-copilot-studio/mcp-add-existing-server-to-agent

---

🧪 Testing

Health check

curl https://your-app.azurewebsites.net/health

List tools

curl https://your-app.azurewebsites.net/tools