How I use Copilot CLI to build Shopify apps on Azure

A hands-on workflow using plans, skills, sub-agents, /fleet, MCP, code generation, and independent review

Author

Dr. Aruna Kishore Veleti
August 12, 2026

In recent past I have been working building enterprise scale Shopify solutions. In this post my hands-on article about how I use GitHub Copilot CLI while building Shopify applications on Azure.

I wanted to go beyond the usual “ask AI to generate code” examples.

The article follows a realistic TypeScript headless middleware application that:

  • Connects storefront experiences to Shopify

  • Manages public, private, and Admin API token boundaries

  • Authenticates and deduplicates Shopify webhooks

  • Queues processing through Azure Service Bus

  • Applies enterprise business rules

  • Updates internal APIs and database systems

  • Publishes normalized events to Azure Event Hubs

But the architecture is only half the story.

The main focus is how I use Copilot CLI throughout the development workflow: 

/plan defines the architecture and boundaries.
Instructions preserve project standards.
Skills capture repeatable procedures.
Specialist agents investigate focused concerns.
/fleet parallelizes independent implementation.
MCP provides approved external context.
Validation and code review protect the result.

One lesson stood out while writing the article:

Copilot’s value isn’t limited to generating a webhook handler. Its real value is helping coordinate how the feature is planned, divided, implemented, validated, and reviewed.

The developer still owns the architecture, security boundaries, production behavior, and final decision. Copilot CLI provides a more structured way to execute that responsibility.

If you’re using Copilot CLI, I’d be interested to know: are you mainly using it for code generation, or have you started experimenting with skills, agents, /fleet, and MCP?

Reply and let me know.