DEV Community

Kabir Adm
Kabir Adm

Posted on

How I Built a Full Xiangqi Game in a Week Using GitHub Copilot (Instead of 3 Months with a small Team)

Three years ago, building a Xiangqi (Chinese Chess) game required a team and months of development.

Today, I rebuilt it almost entirely by myself — with the help of GitHub Copilot.

You can try the game here: Cờ Tướng

This article is not about hype. It's about the real workflow changes, practical lessons, and how AI fundamentally changed how I build software.


The Old Way: 3 Months, 3 People

Previously, building a Xiangqi game required:

  • 2 developers
  • 1 designer
  • ~3 months of work

Scope included

Core Screens

  • Login page
  • Register page
  • Lobby
  • Gameplay screen
  • Table management system

Designer workload (~1 month)

The designer had to create:

  • Board textures
  • Piece designs
  • UI layout system
  • Lobby visuals
  • Buttons, panels, effects

Game UI design is time-consuming because everything must feel immersive and consistent.


Development workload (~2 months)

The developers focused on:

1. UI implementation

Building pixel-perfect UI from design files.

2. Multiplayer system

  • Room creation
  • Joining tables
  • Synchronizing moves
  • Handling reconnects

3. AI engine

This was the hardest part.

We had to implement:

  • Move generation
  • Position evaluation
  • Difficulty levels
  • Avoid obvious blunders

4. State management & bug fixing

Handling all edge cases:

  • illegal moves
  • disconnections
  • game ending logic
  • synchronization errors

Even releasing a basic playable version took around 3 months.


The New Way: Just Me and GitHub Copilot

Recently, I decided to rebuild the game completely from scratch using GitHub Copilot.

No designer.
No second developer.

Just:

  • Me
  • Copilot

My role shifted from writing code → directing code.

My workflow became:

  1. Think of an idea
  2. Write a prompt
  3. Review Copilot's code
  4. Fix or refine with prompts
  5. Repeat

This dramatically reduced development time.


How I Actually Use Copilot

Copilot works best when treated like a junior developer that is extremely fast, but needs direction.

Not magic. Not perfect. But incredibly productive.

The key is how you prompt.


Lesson 1: Build UI One Component at a Time

Bad prompt:

Create a full lobby UI for my game

Result: messy, generic, unusable.

Good prompt:

Create a TableManagementPage with:

  • wooden texture background
  • table list with pagination
  • each table has name, time limit, turn limit
  • user can click to join a table
  • use React and Css module

This produces much higher-quality results.

Why?

Because Copilot performs better when the scope is clear and focused.


Lesson 2: Let Copilot Use Your Existing Architecture

Once your frontend structure exists, backend becomes extremely fast.

Example prompt:

Create API endpoints for TableManagementPage

Copilot automatically generates:

  • GET /api/tables
  • POST /api/tables
  • JOIN table endpoint
  • validation logic
  • models

Because it already understands your codebase.

You don't need to describe everything.

Context is enough.


Lesson 3: Designer Work Reduced by 90%

Before:

Designer needed ~1 month to create full UI.

Now:

I simply prompt:

Create a wooden texture Xiangqi board UI with traditional Chinese style

Then refine:

Make the board darker
Add hover highlight
Add move animation

Copilot generates usable UI instantly.

I only adjust small details.


Lesson 4: AI Handles Most Boilerplate

Copilot excels at writing:

  • React components
  • API endpoints
  • state management
  • validation logic
  • repetitive UI code

This eliminates most of the tedious work.

I focus only on:

  • architecture
  • game logic
  • final decisions

Lesson 5: Development Becomes Exponentially Faster

Old process:

  • Think → Code → Debug → Repeat

New process:

  • Think → Prompt → Review → Done

The difference is massive.

Tasks that took hours now take minutes.


Example: Creating Table Management System

My actual workflow:

Step 1: Prompt Copilot

Create TableManagementPage with pagination, wooden style, join button

Step 2: Copilot generates full React component

Step 3: Prompt backend

Create API for TableManagementPage

Step 4: Copilot generates endpoints

Step 5: Connect frontend & backend

Done.

Total time: less than 1 hour.

Previously: 1–2 days.


My New Role: From Developer → Director

I no longer spend most of my time typing code.

I spend most of my time:

  • thinking
  • designing architecture
  • reviewing AI output

Copilot handles implementation.

This is a fundamental shift.


What Copilot Is NOT Good At

Copilot still struggles with:

  • complex game algorithms
  • architecture decisions
  • long-term consistency
  • optimizing AI engine logic

You still need engineering experience.

Copilot accelerates developers.
It does not replace them.


What Used to Take a Team, Now Takes One Developer

Before:

  • 2 developers
  • 1 designer
  • 3 months

Now:

  • 1 developer
  • GitHub Copilot
  • dramatically less time

This is not theoretical.

This is production reality.


Final Thoughts

GitHub Copilot completely changed how I build software.

It removes most repetitive work and allows me to focus on what matters.

The future of development is not AI replacing developers.

It's developers using AI to build faster than ever.

If you're not using AI in your workflow yet, you're already behind.


If you have questions or want to discuss building games with AI, feel free to comment.

You can try the game here:
https://cotuong.club/

Top comments (0)