Inspiration Section:
- Personal story about the frustrating manual image collection process
- Looking for a simple solution to have Printing store, to print the generative AI images on Poster, T-shirt, mugs etc.
- Connection to Kiro.dev and spec-driven development approach
What it does:
- It allow users to Pass their favorite website url to collect images from it.
- It respect the robot.txt, nofollow for the website.
- It does 3 things: -- Users able to build Collage from images they found from the website. -- Users able to download images from the website(currently I restrict to max 50 and dept 1) for performance and simplicity. -- Users able to select few images, add text description and generate AI Blended images. For this I am using Google Generatvie AI. Currently it's mocked, however imagan model is configuriable.
How we built it:
- Detailed architecture breakdown
- Step-by-step Kiro development workflow
- Technology stack with justifications
- Emphasis on structured AI-assisted development
Challenges:
- In the start Kiro started with spec development mode, that generate specs, design and tasks, but the task list was too big to follow along.
- Kiro keep loosing the token created while development and needs to wait for few days to resume work.
- Real-world problems like CORS, security, memory management
- Not able to automate the deployment on Vercel, and automate with github actions.
Accomplishments:
- Organized into Technical, Architecture, Development Process, and Ethical categories
- Specific metrics (85% test coverage, sub-2-second load times, zero security vulnerabilities)
- Demonstrates both technical excellence and responsible development
- I also refactor the code as per SOLID design principle, Kiro adapt and refactor the code very easily
What we learned:
- Specification driven development vs Vibe Codig by Kiro IDE
- Technical lessons about web crawling and security
- User experience and accessibility learnings
- About Project Steering documents
- Kiro Agent hooks feature, really like this feature.
What's next:
- Could integrate with Image/Video Generating API, and seed these images to get something awesome
- Platform expansion possibilities, to scale from 50 images to thousands
- Can make it a realtime ecommerce store.
Built With
- axios
- cheerio
- material-ui
- node.js/express
- typescript
Log in or sign up for Devpost to join the conversation.