Learn
Topics Overview Introduction to agent-first development Your first agent session Reviewing and controlling agent changes Agent sessions and where agents run Debugging agent work Build your first app with agent mode Why customization matters Instructions Skills Custom agent Hooks Prompt files Customization Features Explained Customization Features In Practice Using tools with agents Extending agents with MCP servers Agent plugins Using third-party agents in VS Code Getting started with Foundry Toolkit in Visual Studio Code Exploring models with Model Catalog Building a social media content agent with Agent Builder Building a hosted agent with GitHub Copilot and Microsoft Foundry Build and Run Your First Spring Boot App Debug and Inspect a Spring Boot Request Expose Your Java Operations to GitHub Copilot with MCP Let GitHub Copilot Test Your Spring Boot App with PlaywrightOn this page there are 9 sections
Understanding VS Code customization features together
There are quite a few customization options in VS Code.
And its easy to get them confused.
This guide breaks down how features like custom instructions, prompt files, agent skills, custom agents, and hooks all fit together and shows how you can use Copilot itself to understand the differences.
Why this gets confusing
All of these features overlap in some way.
They all influence how AI behaves, but they operate at different levels. Without a clear mental model, its easy to mix them up or use the wrong one for a task.
Instead of trying to memorize definitions, a better approach is to explore them interactively.
Using Copilot to learn the differences
One of the most effective ways to understand these features is to ask Copilot directly.
For example, you can provide a list of features and ask for a short explanation of each. Copilot will generate concise summaries that help you quickly understand what each one does.
[Screenshot of Copilot Chat explaining the differences between VS Code customization features]
This gives you a fast, high-level overview without needing to read through multiple docs.
Building a comparison chart
To go deeper, you can ask Copilot to generate a comparison chart.
[Screenshot of Copilot Chat generating a comparison chart of VS Code customization features]
This chart can include:
- what each feature is
- when to use it
- file types or formats
- scope (user, workspace, etc.)
This turns abstract concepts into something you can scan and reference quickly.
Key differences at a glance
Copilot can also highlight the differences between commonly confused features.
For example:
- Instructions vs Prompt Files
- Instructions vs Custom Agents
- Custom Agents vs Skills
- Hooks vs everything else
These comparisons help clarify how each feature behaves and when it should be used.
Learning through scenarios (quizzes)
Another powerful approach is to quiz yourself.
You can ask Copilot to generate scenario-based questions where you choose the correct feature for a given situation.
For example, if you want AI to always follow a specific coding style, the correct answer would be custom instructions.
This method reinforces understanding much faster than passive reading.
Creating a reference you can reuse
Once you have summaries, comparisons, and examples, you can take it one step further and consolidate everything into a single reference.
For example, you can ask Copilot to generate an HTML file that includes:
- feature explanations
- comparison charts
- key differences
- quizzes
This gives you a reusable resource you can revisit anytime.
Why this approach matters
Instead of switching between documentation pages, youre using Copilot as an active learning tool.
This approach helps you:
- understand concepts faster
- see how features relate to each other
- build your own reference materials
It turns learning into something interactive instead of passive.
This represents a shift from:
reading documentation learning with AI
Youre not just consuming information. Youre generating explanations, comparisons, and examples tailored to your understanding.
Whats next
Now that you understand how these features fit together, the next step is applying them in a real project.
Theres no better way to reinforce this than building something and using each feature in context.