Cursor in 10 seconds

• AI code editor built on top of VS Code
• Reads your entire project, not just one file
• You describe what you want, it writes the code
• Most popular AI editor right now with the largest community
• Alternatives: Claude Code, Windsurf, Antigravity

👉 Download it free at cursor.com

The engineering cycle

How professionals build software. One cycle. 4 steps.

Spec → Design → Code → Test

The 5 Concepts

1. 💬 Chat Talk to Cursor like you'd talk to a colleague. Describe what you need, it writes the code, you review it.

Like ordering at a restaurant. You tell the waiter what you want, the kitchen makes it, you taste it.

2. 📌 Context (@-mentions) Cursor can only help with what it can see. Use @file, @folder, or @codebase to point it at the right parts of your project. Better context in, better code out.

Like shining a flashlight into a dark fridge. Light up the right ingredients and you get a better dish. Miss them and the chef is guessing.

3. 🎼 Composer One prompt, multiple files changed at once. Describe a whole feature and Composer builds it across your entire project in one go.

Like standing at the kitchen counter, telling the chef exactly how much salt, when to flip, and what plate to use. You taste at every step. Nothing hits the table without your nod.

4. 🤖 Agents Hand off an entire task. The agent plans the approach, writes the code, runs it, and fixes errors on its own. You come back and decide what to keep.

Like telling the chef to surprise you. They pick the recipe, grab the ingredients, cook it, taste it, fix the seasoning, and bring out the plate. You just eat.

5. 📏 Rules Files Write your coding standards once in a rules file. Every prompt Cursor handles from that point forward follows those rules automatically.

Like posting the kitchen rules on the wall. Every chef who walks in reads them and cooks the same way. No reminders needed.

The mechanics

The real shift

Your skills still matter. Logic, problem-solving, understanding how things connect. AI doesn't replace that. It amplifies it.

You describe. AI builds. You review.

That's how senior engineers work.

We follow the engineering cycle. Each step uses a concept from Phase 1. Every idea becomes real.

✏️ Spec

Concept used: Chat + Context

Open Cursor. Describe Doorashada: the voting form, the candidate cards, the live dashboard, the state tracker. Point Cursor at reference files using @-mentions.

A clear prompt is a clear spec. The better you describe it, the better the code.

🎨 Design

Tool used: Google Stitch

Turn the spec into a visual design. Layout, colors, spacing, interactions. Stitch generates a working UI from your description.

You don't code the design. You see it first, then build it.

🔌 Connect the design to Cursor (optional)

Concept used: MCP

Add the Stitch MCP server to Cursor so it can read your design directly. Now Cursor sees exactly what you designed, not just your words.

What is MCP? MCP (Model Context Protocol) is a way to connect external tools to your AI editor. Think of it like a USB cable between Cursor and other apps. You plug in Stitch, and Cursor can see your designs. You plug in GitHub, and Cursor can see your repo. One standard, many connections.

MCP Code

Put this in .cursor/mcp.json

{
    "mcpServers": {
      "stitch": {
        "command": "npx",
        "args": ["-y", "@_davideast/stitch-mcp", "proxy"],
        "env": {
          "STITCH_API_KEY": "YOUR-API-KEY"
        }
      }
    }
  }
  

📏 Set the Rules

Concept used: Rules files

Before writing a single line of code, create the .cursorrules file. Add three engineering principles:

These are engineering principles, not Cursor features. We add them to the rules file so AI follows the same standards we do.

Rules to Paste

KISS — Write the simplest solution that works, don't over-engineer.
DRY — Extract repeated logic into reusable functions, never duplicate code.
SOLID — One responsibility per module, depend on abstractions, keep interfaces small.

🗺️ Plan

Mode used: Cursor Chat (Plan mode)

Before building, ask Cursor to plan the approach. What files do we need? What components? What's the folder structure? Review the plan, adjust, then move to build.

Planning before building. That's the engineering way.

🔨 Build with Composer

Concept used: Composer

Composer builds Doorashada. The voting form, candidate cards, dashboard, state tracker. Multiple files, one prompt, all at once.

Speed without standards produces messy code. That's why we set the rules first.

🤖 Build with Agent

Concept used: Agent

Agent takes over a full feature. The live counter, animated bars, confetti celebration. It plans, builds, and runs.

Composer = you direct step by step. Agent = you delegate the whole task.

🔍 Review

Read every line AI generated. Accept what's good. Reject what's not. Refine what's close.

Senior engineers spend 40% of their day doing this. Not writing code. Reading code. Making decisions about code.

🧩 Test

Run Doorashada. Cast a vote. Watch the dashboard update. Try to break it.

AI does not test its own code. That is your job.

5 concepts you can use today

1. 💬 Chat : describe what you want
2. 📌 Context : point AI at the right files
3. 🎼 Composer : build across multiple files
4. 📏 Rules files : set your standards
5. 🤖 Agents : delegate full tasks

3 principles that make your code professional

1. KISS : keep it simple
2. Single Responsibility : one function, one job
3. DRY : don't repeat yourself

1 cycle that works everywhere

Spec → Design → Implement → Review → Test

Works with Cursor. Works without Cursor. Works with any tool, any language, any team.

The Promise, Delivered

You came in with coding skills. You're leaving with a superpower.

AI handles the typing. You handle the thinking.

You now know how to:

• Direct AI with clear prompts
• Set engineering standards that AI follows
• Follow the build cycle professionals use every day