What if your AI agent could instantly connect to Salesforce, SAP, Google Drive, and over 1,400 apps and services? With Azure’s API Center, you can expose Logic App Connectors as MCP Servers—giving your agents the power to interact with real systems securely.
In this step-by-step guide, I’ll walk you through how I turned a simple Logic App into an MCP Server, registered it in API Center, and connected it to GitHub Copilot inside VS Code.
Logic App MCP Process Flow
Take a look at the related video below.
Step 1: Create a New Standard Logic App
The process begins inside the Azure Portal. You’ll need to create a new Logic App (Standard Edition) since API Center currently requires a fresh instance.
-
Select Workflow Service Plan
-
Choose your region (I used Central US)
-
Name your app (mine was
APIM Demo) -
Review & Create
Step 2: Set Up API Center
Next, create a new API Center resource. Think of API Center as your hub for organizing APIs and MCP Servers.
-
Choose Free Tier (the Standard edition is quite expensive)
-
Register your Logic App inside API Center
-
Use the MCP Preview feature on the left-hand menu
Azure API Center MCP Setting
Step 3: Register an MCP Server with a Connector
Once API Center is ready, register your new MCP Server:
1. Select your Logic App from the dropdown
2. Add a connector (Google Drive in this demo)
3. Authenticate with Google to complete setup
This step auto-creates a Logic App workflow with the required Request and Response actions, along with Easy Auth for secure authentication.
Setup Azure Logic App in API Center
Step 4: Connect the MCP Server to VS Code
With your MCP Server setup, it is time to connect it to VS Code with GitHub Copilot:
-
In VS Code, open the terminal
-
Run
MCP Add Server -
Enter your MCP endpoint URL (
https://<logicappdomain>/api/mcp) -
Authenticate with Easy Auth
Just like that, your Logic App connector becomes accessible to your AI agent inside VS Code.
Add MCP Server in VS Code
Step 5: Automate Real Workflows (Example: Google Drive Archiving)
In my demo, I had Copilot generate a C# API project plan, but instead of saving it locally, I instructed the agent to send the output to Google Drive via my MCP Server.
This gave me:
-
A consistent archive of coding sessions
-
Ability to compare outputs across models
-
Security (credentials stay with the MCP Server, not the local machine)
Why This Matters
Exposing Logic Apps as MCP Servers unlocks incredible flexibility:
-
Scale instantly with over 1,400 Microsoft-managed connectors
-
Securely authenticate via Easy Auth
-
Empower AI agents to interact with real systems like SharePoint, SAP, or Salesforce, all with complete tracking and management through Logic Apps.
At the end of the day, this is all about transforming how agents interact with enterprise systems.
Conclusion: Start Building Your Own MCP Servers
With just a few steps, you can turn any Logic App connector into an MCP Server using API Center. From archiving code in Google Drive to integrating SAP data into AI workflows, the possibilities are endless.
👉 Want to go deeper? Check out my other posts on:
-
Getting Started with GitHub Copilot in VS Code – https://www.stephenwthomas.com/azure-integration-thoughts/vibe-coding-build-a-c-api-app-with-ai-agents-in-vs-code-no-coding-needed/
-
Using Azure Logic Apps as AI Agents – https://www.stephenwthomas.com/azure-integration-thoughts/step-by-step-guide-to-azure-logic-apps-agent-loop/
Stay tuned for more hands-on demos as we explore how MCP and Logic Apps are redefining integration in the AI era.
Recent Comments