JavaRush/Courses/Claude Code SELF/Workspace: IDEs and plugins

Workspace: IDEs and plugins

Open

1. One panel is not enough

Oliver sits down next to him, nods at the browser where Tommy still has yesterday’s panel open, and asks in a friendly way:

“So, how do you plan to live from here?”

“And what’s wrong with it, Uncle Oliver?” Tommy even gets a little cocky, stands his ground for yesterday’s panel he managed to grow attached to in a single day. “It worked great yesterday. I gave the AI agent a task — it made the file itself, launched the game. Basically, I’ve already got the hang of it. What else do I need?”

“I see,” Oliver says. “So you’re sitting in a little chat window.”

“Why a chat window right away?”

“What else would you call it? A little window in the browser, you type into it, it answers you.” Oliver says it without gloating, more like wearily. “I’ve seen plenty of those chats. Asked — got a wall of text — copied it — pasted it — it didn’t work — went back to ask again. Pretty, but it has nothing to do with real work.”

Tommy opens his mouth to argue — and catches himself. Yesterday he was explaining exactly this to Alex, only from the other side.

“Uncle Oliver, but it doesn’t answer. It does things. It writes files by itself.”

“There.” Oliver raises a finger. “That’s exactly the difference, and it’s good that you caught it. An AI that answers with text is one thing. Claude Code, which climbs into the project itself and works with its hands, is another. You installed it yourself yesterday, by the way — it was the first item on the list.”

“Well, yeah.”

“So then, if it works inside the project — you should be sitting where the project is too. Not in a little window off to the side, but inside. Seeing all the files at once, scrolling code back and forth, with the terminal at hand. Otherwise it’ll churn through half a dozen files for you, and you’ll be hunting them down one by one, guessing what changed where.”

“So how should it be?” Tommy asks.

“In a code editor. It’s a special program for writing programs, called an IDE. You should already have it installed, by the way — I gave it to you yesterday in the list of programs to install.”

Tommy blinks.

“It’s installed. But I haven’t opened it even once.”

“So open it,” Oliver says. “Imagine a carpenter’s workbench: tools hang on the wall to the left, the workpiece is in front of you, a drawer is under your hand — everything is visible, everything is at hand. A code editor is the same: on the left is a list of all the project’s files, in the center is the open code, and at the bottom is the terminal, the very same one I showed you yesterday. Not commands in the dark, but the whole workspace at once.”

Diagram of a code editor: the project's folder and file tree on the left, open code with tabs in the center, and a terminal at the bottom

Tommy thinks it over — and runs into one thing:

“Listen, Uncle Oliver, so will the AI agent move there too? Or does it only live in the browser?”

“That’s what we’ll check.” Oliver smiles. “Let’s relocate it into the IDE.”

2. What you already have installed

“The two of us won’t have to install anything,” Oliver says, and a shadow of yesterday’s notepad creeps into his voice. “You were supposed to install everything yesterday. But since you’re going to use it, let’s at least figure out what has taken up residence there.”

“First. The editor itself — IntelliJ IDEA. A workspace for any project.”

Tommy opens it — and a large window with three zones unfolds in front of him. Oliver points a finger:

IntelliJ IDEA window with three zones outlined: the project file tree on the left, the editor with an open file in the center, and the terminal at the bottom with the claude command typed in

“There, look. On the left is the file tree. In the center is the code editor, here is the open file. At the bottom is the terminal panel. See? The terminal isn’t in a separate window, it just became one of the strips on the screen. This whole thing together is your project’s space at once.”

Tommy zones out for a second: yesterday there was a little slit of a window in the browser, and now — a whole room where everything is laid out. And then it hits him:

“Uncle Oliver, and where’s my work? I spent the whole morning digging around in the browser. Do I have to redo it all now, or what?”

“Well, open it and see.” Oliver shrugs.

Tommy opens it — and freezes. In the tree on the left are his own files. The morning’s spec, and yesterday’s toys. Exactly what he was doing on the website, in the browser, on the other side of town from this laptop.

IntelliJ IDEA with the JavaRush plugin: an arrow points to the Tasks panel button on the left, where the Claude Code SELF course task list is open

More on the magic of moving a solved task’s code over — later in the lecture.

“Uncle Oliver. How did it get here?”

Oliver leans in, looks at the screen for a couple of seconds.

“No idea,” he says at last. “Magic, probably. Fine, you’ll figure it out later.”

And this is probably the first time Tommy sees Oliver not knowing something.

“Second. The editor takes plugins. An add-on to a program is called a plugin — it’s like a drill attachment: the drill itself is one thing, but with attachments you give it new abilities. I’ll explain one of them to you now — that one’s my department. There’s something else installed there too, but that’s not for me, I don’t get into that stuff.”

3. The Claude Code plugin is a bridge

“The plugin that is my department is Claude Code. Official, from the same folks who make the agent itself. And here,” Oliver says, pausing, “there is an important point. This plugin DOES NOT contain the AI agent itself.”

Tommy does not understand right away:

“Wait. So what does it contain then, if not the agent? Why is it there?”

“It is a bridge. There are two different things on your computer: the AI agent itself and the wire to it from the editor. The agent lives separately, on its own. And the plugin just runs a wire from the IDE to it. This is not a second Claude Code and not a second installation — it is the outlet that an already existing device plugs into.”

“And what if the agent weren’t there?”

“Then the plugin simply wouldn’t work.” Oliver shrugs. “The IDE would say something like ‘cannot launch Claude Code,’ and that would be it. There’s a cable, but nothing on the other end. An outlet without a device. The order here is ironclad: first the agent itself, then the wire to it. In your case, by the looks of it, everything landed correctly.”

That’s the check that everything is in place: the Claude Code panel is available in IDEA, and the editor doesn’t complain about a missing agent. If it complains, then it’s not about the plugin — you have to fix the agent installation itself.

The Claude Code plugin is a bridge, not a second agent: the IDE connects through the plugin to the agent, which is installed separately; the agent is installed first, then the plugin; if there is no agent, the plugin won’t start, and it’s the agent installation that needs fixing, not the plugin The plugin is a bridge to the agent, not a second agent Everything is in place IDE code editor Claude Code plugin a cable from the editor no agent inside Agent installed separately Order: the agent first, then the plugin No agent IDE Claude Code plugin empty “Can’t start Claude Code” — fix the agent installation, not the plugin

“So yesterday they installed the tool for me,” Tommy says, figuring it out, “and here they just ran a wire to it from the editor. Not two AI agents, then. The same one, only now it’s at hand.”

“Exactly.” Oliver gestures over the assembled picture. “Well, there. The editor is there, the agent is wired in. And note — nothing had to be installed again. Now just run claude in the terminal (inside IDEA) and that’s it. Thanks to the Claude Code plugin, it will get access to the current project’s files on its own.”

“Listen, Uncle Oliver, and here… not bad, you can work here,” Tommy lets out, and he himself hears that not a trace remains of the morning’s “I’ll make a fool of myself here.”

“Great, get to it. If you have questions — call me.”

4. The JavaRush plugin — a training ground

When Oliver leaves, the first thing Tommy does is move the Terminal panel to the right — like in the WebIDE: it’s both more familiar and more convenient. Then he follows Oliver’s advice and types claude in the terminal. A couple of seconds later Claude Code starts up and shows a field for typing messages.

“Alex, are you there?” Tommy writes into the agent panel right inside the editor.

“Of course! Where else would I go?” Alex says reassuringly. “Why did you call?”

“I’ve moved into IDEA. And I’ve got two questions right away. First: where did my files come from here? I wrote them in the browser, and here they are. Oliver says he doesn’t know.”

“And he shouldn’t,” Alex replies. “That’s our thing. You’ve got a second plugin installed here — JavaRush, the training ground. Everything travels through it.”

“What do you mean, travels?”

“Look. When you submit a solution for checking — no matter from where, from the browser or from here — it goes to the server and stays there. And when you open the task in a new place, it gets pulled from there. You started in the browser, continued in the editor — it’s one and the same work, it just doesn’t live on your laptop but on the server. Nothing has to be carried by hand.”

Tommy is quiet for a second, digesting it. He had been getting ready to redo everything from the morning all over again.

“Handy,” he says at last. “And the second question: what else is this JavaRush plugin for?”

“For the main thing. To build your skills, you need small tasks — lots of them, all different. Not ‘build a big project,’ but ‘do this little thing and submit it for checking.’ The plugin brings them in: a list of assignments right in IDEA, you take them one by one and submit them.”

IntelliJ IDEA with the JavaRush plugin: the task list on the left, the landing page spec file open in the center, the Condition panel with the task statement and requirements on the right, and Validate, Run and other buttons on top

The task list shows up on the left. You take a task you haven’t solved yet and open it. Every task has a description and a list of requirements. Plus a starting set of files.

You read the description, the requirements, and then you solve the task. And when you decide the task is ready, send it to the server for checking. There are buttons at the top for that.

“Hold on.” Tommy squints. “The server checks it itself? Without a human?”

“By itself. It looks at what came out and says whether it counts or not.”

“Yeah, that’s really interesting,” Tommy agrees. “It’s easier to get smarter by collecting a lot of small bumps than one big one.”

5. Two workspaces, and that’s normal

“Just one thing we should say out loud right away,” Alex adds, “or you’ll get confused.”

“Now you have two places where you can work with the AI agent. The first is the panel in the browser, the very one where you played Arkanoid yesterday. The second is this one, the grown-up version: editor, plugins, terminal. And the same AI agent works in both.”

“The same one, exactly one to one?”

“Not quite,” Alex answers honestly. “The panel in the browser is a lightweight version. A trainer: everything you need to learn and solve tasks is there, and nothing extra. The one installed on your machine can do noticeably more — work with any of your files, not just one study folder, run anything, hook external tools up to itself. We’ll get to that, but not today.”

“Got it. A trainer and the real thing.” Tommy nods. “So where should I sit, then?”

“For now, solve the tasks where you’re used to — in the browser,” Alex says. “It’s simple there and there’s nothing to set up. And take your time getting comfortable with this environment, no stress: once you’ve figured it out, you’ll move over for good. Nobody’s rushing you.”

“Beautiful,” Tommy says. The old place hasn’t gone anywhere, and the new one can be settled into at his own pace — that suits him completely. “Then show me what’s next.”

“Next is on you,” Alex replies. “Your spec draft, by the way, is still sitting there with placeholders.”

Tommy winces. The placeholders are the spots where Alex made things up at random, because there were no answers. And there are still no answers now: Tony said “I’ll check with the client” and vanished into meetings, the client is silent, and the week is ticking away.

And then it hits Tommy.

Who exactly said he had to wait? He came up with the product himself — a smart cat feeder: it feeds on a schedule, watches through a camera, and the owner sees everything on their phone. A funny, easy-to-grasp thing. He can figure out on his own how to present something like that — it’s not rocket science.

“Listen,” he says to Alex, “how about I decide it myself. Everything they haven’t answered. I’ll do it my way and show them. If they don’t like it, we’ll redo it, that’s quick. At least I won’t be sitting around waiting for the weather to change.”

“Reasonable,” Alex replies. “Just mark that these are your decisions, not approved ones. So that later nobody argues about who came up with it.”

1
Task
Claude Code SELF,  2 level,  2 lecture
Locked
Nobody decides — so you decide 🐱
Nobody decides — so you decide 🐱
Комментарии
  • популярные
  • новые
  • старые
Для того, чтобы оставить комментарий Вы должны sign in
This page has no comments yet