🎯 What You'll Learn 👥 Target Audience 🤖 Copilot Family 📚 Course Structure
Learn to direct and orchestrate AI coding agents from a single desktop app.
Think of the GitHub Copilot app as a desktop cockpit for agentic coding work. Here, agentic means AI agents can plan and take actions for you, while you still supervise what they do. The app brings together sessions, plans, diffs, tests, browser previews, AI chats, issues, pull requests, and more so you can guide that work without bouncing between multiple tools.
This course treats the app as a place to guide and review work, not a magic code button. You'll practice choosing context, picking a session mode, checking evidence, and deciding when automation is appropriate to use.
By the end of the course, you'll be able to:
- Install, sign in and set up the GitHub Copilot app
- Start sessions from prompts, issues, and pull requests
- Explain Interactive, Plan, and Autopilot modes
- Use worktree-backed sessions without colliding with your main branch
- Attach and manage agent context
- Review diffs, run tests, preview a web app, and validate changes
- Use My work view for issues, PRs, review comments, and failing checks
- Understand where settings, instructions, skills, custom agents, MCP servers, plugins, canvases, and automations fit
The main sample used throughout the course can be found at:
samples/book-app-web
This course is designed for:
- Developers who want to orchestrate, guide, and review agent-driven coding work
- Students and self-taught learners who want a guided path
- Teams evaluating how to keep humans in control while agents do more work
- Copilot CLI or IDE Copilot users who want to understand where the desktop app fits
No agentic development experience is required. Basic GitHub, Git, and JavaScript project familiarity will help. The sample app is a small React/Vite project, so basic npm command familiarity helps in the development chapters. Use the current Node.js LTS for samples/book-app-web.
The GitHub Copilot app works with a Copilot plan or with your own model provider. Business and Enterprise accounts need the GitHub Copilot app policy left enabled. That policy is separate from the Copilot CLI policy.
| Product | Where it runs | Best for |
|---|---|---|
| GitHub Copilot app (this course) | Desktop app | Supervising multi-agent sessions, plans, diffs, browser validation, PRs, canvases, and automations |
| GitHub Copilot in IDEs | VS Code, Visual Studio, JetBrains, and other editors | Agents, inline suggestions, chat, and editor-centered coding |
| GitHub Copilot CLI | Terminal | Terminal-native agent work and command-line workflows |
| Copilot cloud agent | GitHub-hosted environment | Background work on issues and cloud sessions when enabled |
This course focuses on the GitHub Copilot app. Along the way, you'll see how it connects to GitHub, local tools, browser previews, terminal output, and cloud capabilities when available.
| Chapter | Title | What learners do |
|---|---|---|
| 00 | Setup | Prepare the course environment |
| 01 | Tour the App | Learn why you'd use the app, then tour key features: UI, Chats, settings, sessions, modes, and model controls |
| 02 | Sessions, Worktrees, and Context | Start isolated sessions and use @, #, and / for context |
| 03 | Development and GitHub Workflows | Review, debug, test, and preview a change, then move it through My work, issues, PRs, review comments, checks, and guided fixes |
| 04 | Skills and Custom Agents | Update a review skill, create a read-only custom agent, and validate one skill-guided improvement |
| 05 | MCP Servers and Plugins | Retrieve documentation through an MCP server and use a plugin's skill for a focused recommendation |
| 06 | Canvases | Run /create-canvas for a visual session board to keep the plan, progress and validation evidence visible |
| 07 | Automations | Start with a manual open-work summary, then learn schedules and optional cloud automations |
Each chapter follows the same beginner-friendly pattern:
- An introduction - why the topic matters
- A supporting real-world analogy
- Core agent-development concepts
- Hands-on examples using
samples/book-app-web - Key takeaways, an assignment, and additional resources
Note
When a chapter shows a model response, remember that model output varies due to the non-deterministic nature of LLMs. Your app version, model, reasoning setting, repository context, and enabled tools can also change the structure of the response.
- GitHub Copilot app overview
- GitHub Copilot app videos
- Getting started with the app
- Working with sessions
- Issues and pull requests
- Using automations
- Working with canvas extensions
- Customizing the GitHub Copilot app
- Public app repository
- GitHub Copilot app changelog
This project is licensed under the terms of the MIT open source license. Please refer to the LICENSE file for the full terms.