Exploring IDE Agents

00:00 IDE agents live inside your code editor and work alongside you in real time. They suggest edits inline, show visual diffs, and let you accept or reject changes without ever leaving your editing environment.

00:14 They come in two flavors. There are AI-native IDEs and IDE integrations. AI-native IDEs were built from the ground up around AI capabilities, while IDE integrations add agent features to the editors you already use.

00:29 That said, the choice is developer-specific, so try a few and see which style suits how you actually work.

00:37 AI-native IDEs are tools like Cursor, Windsurf, and Kiro. Some of them, like Kiro, support a spec-driven workflow. You describe what you want up front, and the agent works through a plan before ever touching a single line of code.

00:53 IDE integrations include the GitHub Copilot extension and the Claude Code extension for VS Code. So let’s see GitHub Copilot in action. Here you can see Visual Studio Code.

01:06 One way you can interact with Copilot is by typing a function, a comment, or a variable name. For instance, you could type the definition of a function to calculate the area.

01:17 Copilot will show a suggestion in gray text, which you can accept or reject. In this case, it assumed that the area was that of a rectangle. But remember, you’re always in control.

01:30 Suggestions are gray text until you decide to accept them. Another way you can interact with Copilot is by using the inline chat. So let’s imagine that you have to fix a bug in this code that you wrote.

01:44 Your task is to comply with your company’s privacy policy and remove inactive users who haven’t logged in for more than five years. So you wrote this code, and it isn’t working properly.

01:56 So you press Ctrl + I to open the inline chat, and you ask Copilot if it can fix this bug and explain in a comment why the current implementation can skip users.

02:07 So Copilot will take some time to understand what is happening, and then it’ll suggest a fix and explain why the current code isn’t working. In this case, it said that removing items while iterating shifts the remaining users left, so a user can be skipped by the loop.

02:25 So here again, you can decide to keep or undo those changes.

02:32 You’ve just seen how the inline chat can help you fix a specific bug. But what if the task is larger and involves multiple files? This is where the Copilot chat panel becomes useful.

02:44 Press Ctrl + Alt + I to open it. The Copilot chat offers three modes, Ask, Plan, and Agent.

02:53 Ask mode is useful when you want to ask questions about your project without making changes. For instance, explain briefly what this code does.

03:04 Copilot analyzes the code and gives you an explanation of what the script does.

03:11 Then you have Plan mode.

03:14 Plan mode is useful when you have a larger task and you want Copilot to figure out how it would approach the problem before making any changes. For example, turn the script into a reusable function where the user can specify the number of years of inactivity as an argument.

03:34 Copilot analyzes the project and creates an implementation plan, but doesn’t make the changes yet. You can review the plan here, and once you’re happy with the plan, you can hand it over to Agent mode.

03:48 Agent mode can implement the changes, work across multiple files, and run commands when needed. Every now and then, it will ask for you to accept or reject the changes it’s planning to make.

04:08 Once Copilot is done, you can choose to keep or reject the current changes. The chat panel also lets you choose the model and depending on the model, adjust the thinking effort.

04:22 One important thing to understand is that most IDE agents are cloud-backed, and that means that your code gets sent to external servers for processing.

04:32 If your team has policies about where proprietary code can go, check those before adopting a tool. If your code can’t leave your machine at all, a tool like Continue lets you run models locally instead.

04:46 In the next lesson, you’ll learn more about terminal agents.