🎬 Watch the full walkthrough: Cursor 101 on YouTube
“Everybody can be a builder.” That’s the promise of AI-powered coding tools — and with Cursor, it’s actually true. Here’s how to get started from absolute zero.
There are plenty of videos out there about vibe coding, AI-assisted development, and tools like Cursor, Claude, and Gemini. But most of them have the same problem: they assume you already know the basics. They throw around terms like “scaffold a backend” or “clone a repo” without ever explaining what that actually means.
This post (and the companion video) is different. I’m going to walk you through Cursor — one of the most powerful AI coding tools out there — starting from the very beginning. No jargon. No assumptions.
My name is Sean, also known as Milhouse. I’m a technical program manager, former software engineer, vibe coder, and someone who teaches people how to use this technology in real workplaces and university settings. I’ve been where you are — staring at Cursor for the first time, completely unsure where to start. Let’s fix that.
What is Cursor, exactly?
Cursor is an AI-powered code editor. It’s built on top of a popular developer tool called Visual Studio Code (VS Code), which is essentially just a really smart text editor for writing code. But Cursor takes things a step further by letting you chat with it — you describe what you want to build, and it builds it for you.
You don’t need to know how to write code. You just need to know how to describe what you want. Cursor will either make smart assumptions on your behalf or ask you for more detail. Either way, it handles the heavy lifting.
The one-sentence version: Cursor is like having a brilliant developer sitting next to you — except you’re the one driving, and you communicate in plain English.
Getting started: your first project
When you open Cursor for the first time, you’ll see a simple screen with a few options: open a project, clone a repo, or connect via SSH. For now, forget the last two. All you need is “Open a Project.”
Here’s the workflow I recommend for staying organized:
- In your home directory (Mac or Windows), create a folder called development. This becomes your home base for every project you ever build.
- Pin that folder to your Finder favorites (Mac) or Quick Access (Windows) so you can get to it in two clicks.
- When starting a new project, create a new subfolder inside development — something like youtube-demo or my-first-app.
- Open that folder in Cursor. You now have a clean slate.
Finding your way around the interface
Once you’re inside Cursor, here’s what you’re looking at:
The file explorer (usually on the left)
This is basically a file browser. As Cursor builds things for you, files and folders will appear here. You can click through them to see the code — or just ignore them and let Cursor handle everything. Both approaches are totally valid when you’re starting out.
The agent chat (usually on the right)
This is where the magic happens. Look for a panel that says “New Agent” with a blinking cursor. If you don’t see it, look for a small chat bubble icon in one of the corners — click that to toggle the panel. This is where you type your instructions and have a back-and-forth conversation with the AI.
The mode dropdown
Below the chat window, you’ll find a dropdown with options like Agent, Plan, Debug, and Ask. Here’s the quick breakdown:
| Mode | What it does |
|---|---|
| Agent | Default. Does everything — plan, build, fix. Start here. |
| Plan | Map out the project before writing any code. |
| Debug | Troubleshoot when something breaks. |
| Ask | Ask questions without making any code changes. |
I almost always leave mine on Agent. It’s the jack of all trades — it can build, answer questions, and troubleshoot without switching modes.
Choosing the right AI model
Next to the mode dropdown, you’ll see a model selector. By default it’s set to “Auto,” which usually picks a Cursor-native model. But you can switch to others based on what you’re doing:
- Heavy lifting (complex apps, architecture decisions): Use Sonnet or Opus from Anthropic
- Lighter tasks (quick edits, simple questions): Use Haiku to conserve credits
- Thinking models (the ones with a brain icon next to them): More powerful, but they use more credits — save them for complex troubleshooting
If you don’t see a model you want, click “Add Models” to open Cursor’s settings and toggle things on or off.
Now let’s actually build something
Here’s where it gets fun. To demonstrate how Cursor works, I used it to build a drag-and-drop task management app — basically a digital kanban board with sticky notes you can move between “To Do,” “Doing,” and “Done” columns.
Here’s the prompt I used:
Build me a simple task management app that allows me to drag and drop digital sticky notes between columns: To Do, Doing, and Done. It should be elegant, easy to use, and responsive. Use Python, Next.js, and a PostgreSQL database for data. Should run locally.
Notice what I included in that prompt. I didn’t just say “build me an app.” I specified:
- What it should do — drag and drop sticky notes between columns
- How it should feel — elegant, easy to use, responsive
- What technologies to use — Python, Next.js, PostgreSQL
- Where it should run — locally
Those technology choices matter — and a lot of vibe coding tutorials gloss right over this. If you ever want to move beyond your laptop and deploy something to the internet, you’ll need to understand what your app is built on. That way, when you start asking questions like “Should I use AWS or Heroku?”, you’ll actually know what you’re talking about.
Watch Cursor think (chain of thought)
After you send your prompt, Cursor shows you its “chain of thought” — a live readout of everything it’s figuring out. In my demo, it went through planning, scaffolding the backend, setting up the database schema, building the front-end UI, and configuring dependencies. The whole thing took about five minutes.
Most of this will look like gibberish at first, and that’s completely fine. The files start appearing in your explorer panel, and eventually Cursor says: “I’m done. Here’s what you have. Here’s how to run it.”
When I tested the finished app, it worked exactly as described. I typed in a task, added it as a sticky note, dragged it from “To Do” to “Doing” to “Done.” Start to finish: five minutes, zero lines of code written by me.
Pro tip: If Cursor hits an error while building — and it will sometimes — just let it keep going. It’s designed to self-correct. In my demo, it ran into a Docker issue and automatically put a fallback method in place, no input from me required.
One more useful feature: upload images
You can attach images directly to the Cursor chat. This is incredibly useful for a few situations:
- Error messages — screenshot the error and paste it into the chat. Cursor will diagnose and fix it.
- Visual issues — if something doesn’t look right in your app, show it a screenshot.
- Sketches or wireframes — draw your idea on a whiteboard or napkin, photograph it, upload it, and say “build this.”
It’s a surprisingly powerful feature that most beginners never discover.
The big picture
Cursor can feel intimidating. I was a software engineer, and it still took me a minute to wrap my head around it. But once you get going, it becomes one of the most natural tools you’ll ever use.
The key mindset shift: you’re not learning to code. You’re learning to direct. Think of yourself as the product manager and Cursor as your entire engineering team. Your job is to describe what you want clearly — the more specific the better — and let the AI figure out the rest.
That’s it. That’s Cursor 101.
Ready to see it in action?
Watch the full walkthrough on YouTube where I build the entire sticky note app live in under five minutes — including all the behind-the-scenes stuff other tutorials skip.
Have questions? Drop them in the YouTube comments — I respond to every one. And if there’s a follow-up topic you want covered, let me know. We’re just getting started.
Leave a Reply