English | 日本語
Welcome! This is the hands-on workshop repo for AI Genius Episode 1. You'll work through the full AI-native development loop: writing issues, delegating to Copilot, reviewing generated code, and iterating via PR comments.
- What "AI Native" actually means for developers
- How to write issues that give Copilot the context it needs
- How to assign work to Copilot and observe it in action
- How to review Copilot-generated PRs like a senior developer
- How to iterate via PR comments instead of starting from scratch
- Best practices for collaborating with AI throughout the coding process
- In the optional exercise, how to review changes involving cloud SDKs and credentials with safety in mind
IDEA
└─► GitHub Issue (describe the work)
└─► Assign to Copilot (Copilot agent picks it up)
└─► Work runs in a restricted temporary environment
│ (effective access depends on configuration)
└─► Draft PR is opened (summary + View session)
└─► Human reviews and iterates via PR comments
└─► Merge and ship
You are the tech lead in this workflow. Copilot handles the how. You define the what and why.
- A GitHub account with access to GitHub Copilot
- GitHub Copilot App installed and signed in
- Python 3.10+ installed locally (for the starter app)
- Git installed
-
Fork this repo to your own GitHub account (top-right corner of this page).
-
Clone your fork locally:
git clone https://github.com/YOUR-USERNAME/aigenius-s5-ai-native-coding-workflow.git cd aigenius-s5-ai-native-coding-workflow -
Run the starter app:
cd starter-app pip install -r requirements.txt python app.py add "Deploy the API" --priority high --due 2025-12-31 --tag work python app.py add "Buy coffee" --priority low --tag personal python app.py list python app.py stats
-
Open the GitHub Copilot App and connect it to your forked repo.
-
Complete core Exercises 01–04 in order, starting with
exercises/01-write-an-issue. After completing the core workflow, optionally continue to Optional Exercise 05 for Azure and AI practice.
The repository includes a project Canvas extension that presents the AI Genius S5E1 slide images without switching to PowerPoint.
- Open this repository in GitHub Copilot App.
- Start a session after project extensions have loaded.
- Ask Copilot:
Open the AI Genius Slide Presenter canvas. - Use the on-screen controls, thumbnails, or keyboard shortcuts to present.
The viewer supports previous/next navigation, direct thumbnail selection, slide numbers, and fullscreen. If the host does not allow the browser Fullscreen API, it uses an in-Canvas presentation mode instead.
See presentation/ai-genius-s5e1 for controls and slide update instructions.
The bilingual HTML DX guide explains the central message, six changes in developer experience, product roles, and a review-and-iteration framework for making acceptance decisions. Use its 日本語 / English toggle to switch the content, diagrams, prompts, and exercise links. It is a self-contained, scroll-based artifact that can be opened locally or in a Browser Canvas, independently of the slide presenter.
The bilingual Agent Merge guide explains the feature, app workflow, use cases, and conditions people set through CSS/SVG diagrams and five fictional PR simulations. It is a single-file HTML artifact with no runtime network access or real PR operations. This independent companion introduces a further option without changing the core exercises' human-operated merge workflow.
| Track | Exercise | What you will learn |
|---|---|---|
| Core | Exercise 01 | Write a problem statement, desired behavior, acceptance criteria, constraints, and Definition of Done that an agent can implement |
| Core | Exercise 02 | Delegate an issue to Copilot and observe its investigation, implementation, and testing |
| Core | Exercise 03 | Review a generated PR for correctness, quality, security, dependencies, and test effectiveness |
| Core | Exercise 04 | Request improvements through precise PR comments, re-review the result, and merge manually |
| Optional | Optional Exercise 05 | Extend the completed workflow to Azure Table Storage and Azure OpenAI scenarios |
Completing Exercises 01–04 completes the workshop's core learning path. Optional Exercise 05 is an additional cloud-and-AI challenge for learners who want to continue.
Exercise 01 includes Azure-based Options A and B. Choose Option C or D for a core path that does not require Azure access; Options A and B require an appropriate Azure environment and can be deferred to Optional Exercise 05.
📁 aigenius-s5-ai-native-coding-workflow/
├── README.md / README.ja.md # English and Japanese workshop guides
├── .github/
│ ├── copilot-instructions.md # Canonical Copilot context
│ ├── copilot-instructions.ja.md # Japanese contributor reference
│ ├── extensions/
│ │ └── ai-genius-presenter/ # GitHub Copilot App slide Canvas
│ └── ISSUE_TEMPLATE/
│ ├── feature-request.md # English feature template
│ └── feature-request-ja.md # Japanese feature template
├── exercises/
│ ├── 01-write-an-issue/ # Task: write a well-formed issue (cloud/AI options)
│ ├── 02-assign-to-copilot/ # Task: assign + observe
│ ├── 03-review-a-pr/ # Task: review and comment on a PR
│ ├── 04-iterate/ # Task: iterate via PR comments
│ └── 05-azure-and-ai/ # Optional: Azure + OpenAI extension
├── presentation/
│ ├── ai-genius-s5e1/ # Slide manifest, images, and update guide
│ └── copilot-app-dx/ # Self-contained Japanese/English HTML DX guide
└── starter-app/ # Python CLI task manager to extend
├── app.py # CLI: add, list, complete, edit, delete, stats
├── requirements.txt # click, rich, pytest
└── tests/
├── conftest.py # Shared fixtures (isolated task file)
└── test_tasks.py # 41 tests covering all commands + edge cases
- Write better issues -- your issue IS your prompt. Be specific.
- Review like a senior dev -- AI generates fast, humans verify smart.
- Use
copilot-instructions.md-- give Copilot standing context about your project. - Iterate, don't regenerate -- guide via comments rather than starting from scratch.
- Stay in the loop -- check the session log, understand what Copilot did and why.
Shinya Yanagihara -- Global Black Belt, Microsoft Corporation
Shinya works across Microsoft's developer tooling with GitHub at the centre, helping teams adopt AI-native development practices and driving the organisational culture change that makes them stick.