Posted on Jul 21

Rebuilding Retro Space Invaders with AWS Kiro - Spec-Driven Development in Action!

“Learn continually. There’s always ‘one more thing’ to learn!” — Steve Jobs

Originally built using Amazon Q CLI, this time I used AWS Kiro to recreate the game and it delivered a structured, traceable, and developer-friendly workflow along with markdown requirements, design and implementation tasks!

Hi! I’m Girish, an AWS Community Builder & Cloud Tech Enthusiast, and in this article, I’ll walk you through how I recreated the classic retro game Space Invaders using AWS Kiro, the brand-new AI-powered IDE from AWS announced at #AWSSummitNYCon July 16th, 2025.

You might remember my previous build using Amazon Q CLI, where I generated 200+ lines of working Python + Pygame code in under 10 seconds. While that was amazing, I did not have specs and design for that, rather code was directly created by the AI.

With specs driven development, Kiro goes even further!

With Kiro, I not only got the code and working game, I also got specs!

It created:

Structured requirements
Generated design documentation
Created Implementation task lists
Created traceability between tasks and requirements
Generated working code tasks by tasks
Optionally, it can generate unit tests as well.

Let’s dive into how AWS Kiro transforms GenAI-assisted development.

What Is AWS Kiro?

AWS Kiro, a new AI-powered IDE, announced at #AWSSummitNYC, and as an AWS Community Builder, I had the opportunity to work with it before the public release.

While many tools support code generation, Kiro goes further by enabling spec-driven development. From a single prompt, it can:

Generate high-level requirements
Create design documents
Break the design into implementation task lists
Trace tasks back to the original requirements
Generate working code for each task
Optionally create unit tests

This structured and traceable approach helps build greater trust in GenAI-assisted development.

How to Install it?

Kiro is forked from VS Code and it is installed the similar way. You can download the software from kiro.dev.

Once downloaded and installed, you can sign in using one of the four methods provided.

Game Overview

I used the same game idea as before, a full-featured clone of the retro classic Space Invaders.

Player Controls:

Arrow Keys or A/D to move the ship left/right
Spacebar to shoot bullets
ESC to quit the game

Gameplay Features:

A grid of invaders that moves horizontally and drops when hitting edges.
A player ship at the bottom that can shoot bullets upward.
Collision detection between bullets and invaders.
A score system that gives +10 points per invader destroyed.
Win/Lose conditions:
Win: Destroy all invaders.
Lose: If any invader reaches the player's ship.

Visual Style:

Retro “pixel-style” using simple colored rectangles, mimicking the original arcade aesthetic.

Step-by-Step: From Prompt to Playable Game with Kiro

Open Kiro and Enter a Prompt

Let's start with the same prompt that I used in my previous article to create this space invaders game.

Prompt: Create a simple retro game Space Invaders using Python and Pygame.

Unlike many other GenAI assisted tools, rather than directly creating the code, Kiro generated specs first!

It generated:

Requirements
Design Document
Implementation task list
And then generated the code for each task!

Each task was linked to a requirement, making it easy to trace back any piece of code to its purpose.

Review the Generated Artifacts

Requirements.md

Kiro will generate requirements.md file that has requirements organized in sections in a markdown format.

Here is an example of requirements generated:

After requirements are created, Kiro will proceed to create design to explain the components that will be used for code generation.

Here is an example of design generated:

Tasks.md

Once requirements and design is created, it is time for the implementation plan. Kiro generated the tasks list for the implementation plan.

Here is an example of tasks generated:

These artifacts made me feel like as if this entire solution is crafted by a skilled software engineer and not an AI generated solution.

Let's review the generated code. As you can see in the example below, it seems like a modular code with proper structure and definitions.

Run the Game
Once the file is generated, simply run:

python3 space_invaders.py

And it worked beautifully; just like my previous build, but this time I had documentation and structure to support it.

See the Game in Action

Here is the game in action!

Conclusion

Kiro is another example of how far Generative AI has come from generating documentation to building entire applications. While skills of a software engineers are still required to ensure that code has appropriate exception handling and error management and follow the best security practices, the use of GenAI assisted development will improve the cycle time and will help release the software faster which in turn will help not only meet strategic objectives of the business but will also help validating the demand for experimental features fast!

In this article, I: