Here is a polished Devpost-style hackathon submission for Prompt Code-Arena.
Inspiration
Prompt Code-Arena was inspired by a simple observation: while AI coding assistants have become incredibly powerful, most developers still struggle to write clear, structured prompts. A vague prompt often leads to multiple iterations, wasted API credits, and inconsistent results. We wanted to create a platform where developers can practice prompt engineering like they practice competitive programming—through challenges, instant feedback, and a leaderboard that rewards quality and efficiency.
What it does
Prompt Code-Arena is an interactive platform that helps developers improve their AI prompting skills through coding challenges.
Users can:
- Solve real-world software development tasks using only natural language prompts.
- Compare generated code against challenge requirements.
- Receive AI-powered feedback on prompt quality, clarity, and completeness.
- Earn scores and climb a leaderboard based on prompt effectiveness.
- Explore previous solutions and learn better prompting strategies from the community.
Instead of evaluating coding ability alone, Prompt Code-Arena evaluates the ability to communicate intent effectively to AI models.
How we built it
We built Prompt Code-Arena as a modern full-stack web application.
Frontend
- Next.js
- React
- Tailwind CSS
- Responsive UI optimized for desktop and mobile
Backend
- API routes for challenge management
- AI-powered prompt evaluation
- Prompt scoring pipeline
AI
- Large Language Models to generate code from user prompts
- Automated evaluation comparing generated output with expected requirements
- AI-generated feedback highlighting strengths, missing constraints, and suggestions for improvement
Deployment
- Vercel for seamless hosting and continuous deployment
- GitHub for version control and collaboration
GitHub Repository: https://github.com/karun99/proto-collection/tree/main/prompt-code/prompt-code
Live Demo: https://prompt-code-one.vercel.app/
Challenges we ran into
- Designing a scoring system that fairly evaluates prompts instead of simply grading generated code.
- Making AI feedback actionable rather than generic.
- Handling variations in LLM outputs while maintaining consistent evaluation.
- Creating challenges that are realistic, engaging, and progressively more difficult.
- Balancing prompt creativity with measurable performance metrics.
Accomplishments that we're proud of
- Built an end-to-end platform that transforms prompt engineering into a competitive learning experience.
- Successfully integrated AI-based prompt evaluation with meaningful feedback.
- Created an intuitive and responsive user experience.
- Developed a scalable architecture that can support additional challenge categories and AI models.
- Demonstrated how prompt engineering can be learned through practice rather than trial and error.
What we learned
Throughout the project, we learned that:
- Prompt engineering is becoming a critical software development skill.
- Small improvements in prompt structure can dramatically improve AI-generated code quality.
- Building reliable AI evaluation systems requires balancing deterministic checks with model-based reasoning.
- Great developer experiences come from fast feedback, clear scoring, and opportunities for iteration.
- Gamification is an effective way to encourage continuous learning in AI-assisted development.
What's next for Prompt Code-Arena
Our roadmap includes:
- Multiplayer prompt battles where developers compete head-to-head.
- Support for multiple AI providers and models.
- Team competitions and organization-wide leaderboards.
- Difficulty-based challenge tracks for beginners and advanced users.
- Personalized AI coaching with prompt improvement suggestions.
- Historical analytics to track prompting skill growth over time.
- Community-created challenges and public prompt libraries.
- Integration with IDEs to help developers practice prompting directly within their coding workflow.
Ultimately, we envision Prompt Code-Arena becoming the go-to platform for mastering prompt engineering through hands-on practice and friendly competition.
Built With
- react
- supabase
- typescript
Log in or sign up for Devpost to join the conversation.