Copilot Desktop App

What This Demonstrates

The GitHub Copilot app is a standalone desktop application that brings agent-driven development to any local repository — independent of IDEs or the browser. Sign in with your GitHub account, point it at a local clone, and run full agent sessions with the same capabilities available in VS Code or GitHub.com.

Available on all Copilot plans (Free, Pro, Business, Enterprise) and supports macOS, Windows, and Linux.

Prerequisites

• GitHub Copilot app installed (download from gh.io/copilot-app)
• Authenticated with your GitHub account
• Local clone of the octo-eshop-demo repository

Prompts

Prompt A — Repository Overview

Analyze the monorepo structure in the `services/` directory. Identify all microservices,
their responsibilities, and recommend where a new discount-code service would best fit.
Include which existing services it would need to communicate with and what API endpoints
it should expose.

What to look for:

• A specific recommendation on where the discount-code service fits in services/
• Identification of which services (e.g., cart-service, order-service) would call it
• Suggested API endpoint shapes consistent with the project’s success/data/error/meta response format

Prompt B — Implementation Task

Add a health check endpoint to the user-service that returns service status and database
connection info. The endpoint should live at GET /health and return JSON conforming to
`{ success: boolean; data: { status, db, uptime } }`. Follow the existing project conventions in
`services/user-service/src/`.

What to look for:

• New route registered in services/user-service/src/routes/
• Controller that queries Prisma to verify the database connection
• Response conforming to { success: boolean; data: { status, db, uptime } }
• No changes outside the user-service directory

Prompt C — Cross-Service Flow Analysis

Trace the full order creation flow across microservices. Starting from when a user checks
out in the cart-service, walk through how the order reaches order-service and then
payment-service. Include the async events on Azure Service Bus.

What to look for:

• Step-by-step trace from cart-service checkout → order-service creation → payment-service charge
• Identification of the Azure Service Bus events (order.created, order.paid)
• Any gaps or missing error-handling paths surfaced by the analysis

Follow-Up Prompts

• “Generate a Mermaid sequence diagram for the order creation flow you just traced.”
• “Add integration tests for the health check endpoint using the existing test setup in services/user-service/tests/.”
• “What would need to change in order-service to support order cancellation?”