Build Octocat Supply
with agents.

Standalone community edition. Start with the setup guide. This is an adaptation of Microsoft's MicroHack, not an official Microsoft site. The complete runnable starter is in src/; website/ contains the reading guide. Source, licensing, and adaptations.

MicroHack introduction#

This MicroHack walks you through the Agentic Software Development Lifecycle (SDLC) — plan, implement, test, review, deploy, and operate with agents — using GitHub Copilot, GitHub Actions, and Microsoft Azure, applied end-to-end to a real full-stack application called Octocat Supply.

Rather than treating an AI coding assistant as autocomplete, you practise driving agents across the whole lifecycle: refining requirements from organizational signal, planning multi-file changes, generating and hardening tests, reviewing code, deploying to Azure with Infrastructure as Code, monitoring with the Azure SRE Agent, and standing up agentic workflows that maintain the app over time.

flowchart LR
    A[Plan] --> B[Implement]
    B --> C[Test]
    C --> D[Review]
    D --> E[Deploy to Azure]
    E --> F[Operate / Monitor]
    F --> G[Maintain]
    G -.-> A
    subgraph Agents
      Copilot[GitHub Copilot / Copilot App]
      Actions[GitHub Actions CI/CD]
      SRE[Azure SRE Agent]
      GHAW[GitHub Agentic Workflows]
    end
    Copilot --- A
    Copilot --- B
    Copilot --- C
    Copilot --- D
    Actions --- E
    SRE --- F
    GHAW --- G

This lab is not a full explanation of GitHub Copilot or Azure as technologies. The following are recommended pre-reading to build foundational knowledge:

• GitHub Copilot documentation
• GitHub Actions documentation
• Azure Container Apps
• Azure SRE Agent

What's included. This repository contains the workshop's code and teaching assets: the Octocat Supply application source (src/), the Azure infrastructure scaffold (infra/), helper scripts/, the dev containers (.devcontainer/), the editor config (.vscode/), and the repo custom instructions, custom agents, and workflows (.github/). The Challenge 3 inputs — the product backlog and the mock WorkIQ organizational knowledge (Teams thread, support-ticket digest, stakeholder email) — live under assets/. Fork the repository using your own GitHub account and use your own Azure subscription for the Azure challenges. No accounts, subscriptions, or preconfigured lab environments are supplied.

MicroHack context#

The Octocat team are behind schedule launching their new Octocat Supply application and are exploring Agentic Development practices to help them deliver the remaining features before deploying the full application to their Azure environment. Core functionality is still missing — like the shopping cart and other backlog items — and the testing and monitoring aren't yet up to Octocat standards for a reliable, secure application.

Start from the included application code and use your own GitHub Copilot access to complete the work by progressing through the challenges. The app is a full-stack supply chain platform with a React + Vite + Tailwind frontend and four interchangeable API implementations (TypeScript, C#, Python, Java) over a shared SQLite schema — pick whichever API language you prefer; the challenges are language-agnostic.

Objectives#

After completing this MicroHack you will be able to:

• Use GitHub Copilot to plan and implement features across a full-stack app while keeping changes scoped and consistent.
• Run an agentic loop — plan → implement → test → review — on a backlog item, including refining a requirement from ambient organizational signal.
• Generate and extend tests deliberately, choosing models and context budgets on purpose, and run an agent-assisted code review.
• Deploy to Azure with Infrastructure as Code and a GitHub Actions pipeline authenticated via OIDC — no secrets in source.
• Use the Azure SRE Agent to detect, diagnose, and remediate a fault in a deployed app.
• Stand up GitHub Agentic Workflows to maintain the application over time with clear triggers and guardrails.

MicroHack challenges#

General prerequisites#

This MicroHack has a few but important prerequisites. Completing them before the session keeps the time focused on the differentiated, agentic work rather than setup.

In summary:

• Your own GitHub account, your own fork of this repository, and active GitHub Copilot access with the Chat, agent, CLI, and MCP capabilities required for the exercises enabled for your plan.
• Either GitHub Codespaces (recommended — nothing to install) or local VS Code with the Dev Container plus the GitHub Copilot / Copilot Chat extensions. The container preinstalls Node.js 24, Python 3.13, .NET 10, Java 17.
• For Challenges 5 & 6, Azure access:
  • Your own Azure subscription with a budget and cost alerts.
  • A dedicated Resource Group you create for the workshop in that subscription.
  • Contributor permissions on that scope (plus any roles the Azure SRE Agent needs on the monitored resources).
  • A GitHub ↔ Azure OIDC federated credential you configure for your fork's pipeline authentication.
  • Azure SRE Agent available in your selected region and subscription.

Before you begin, verify your own Copilot access and repository permissions. You can optionally seed backlog issues in your fork. Confirm your Azure subscription, budget, and permissions before Challenge 5; no lab credentials or Azure environment are provided.

Challenges#

• Challenge 1 - Dev environment, Copilot access & repo overview <- Start here
• Challenge 2 - Deliver a new feature (the cart)
• Challenge 3 - Add another feature from the backlog (agentic loop)
• Challenge 4 - Run and extend test coverage, and review code
• Challenge 5 - Deploy into Azure
• Challenge 6 - Monitor the deployed app with the Azure SRE Agent
• Challenge 7 - Implement agentic workflows to maintain the app
• Finish

Solutions - Spoilerwarning#

Do not open these before attempting the challenges — they contain guidance and worked approaches.

• Solution 1 - Dev environment & repo overview
• Solution 2 - Deliver the cart feature
• Solution 3 - Backlog feature (agentic loop)
• Solution 4 - Testing & review
• Solution 5 - Deploy into Azure
• Solution 6 - Monitor with the Azure SRE Agent
• Solution 7 - Agentic workflows

Suggested agenda#

A typical one-day delivery of this MicroHack:

Time	Activity
09:00 – 10:00	Tech Talk — Agentic SDLC overview
10:00 – 12:30	Hacking session — Challenges 1, 2, 3, 4
12:30 – 13:30	Lunch break
13:30 – 15:30	Hacking session — Challenges 5, 6, 7
15:30 – 16:00	Final discussion / Wrap-up

Contributors#

• Francesco Lana (GitHub, Linkedin)