Inspiration

Inspiration Large codebases become hard to understand, risky to modify, and slow to refactor. We wanted a developer-first tool that turns messy repositories into clear, visual structures and allows safe changes through simple natural-language commands instead of manual file hunting.

What it does The platform visualizes an entire repository as interactive dependency graphs and provides an AI-powered command terminal for refactoring, navigation, and analysis. Developers can issue natural-language instructions and instantly see structural changes reflected in real time, reducing guesswork and preventing breakages.

How we built it

  • Frontend: Next.js with ReactFlow for dynamic graph rendering and smooth interactions
  • Backend: NestJS with Prisma and BullMQ for job processing and scalable task queues
  • AI Layer: Gemini 2.5 Flash for code understanding, suggestions, and automated modifications
  • Architecture focused on modular parsing, async processing, and visual state streaming

Challenges we ran into

  • Maintaining real-time visual accuracy while processing large repositories
  • Handling multiple file types and edge-case syntax structures
  • API rate limits and batching strategies for AI requests
  • Synchronizing graph state with backend refactor operations without UI lag

Accomplishments that we’re proud of

  • A fully functional natural-language AI terminal for repository actions
  • Clean, industrial-style UI optimized for large graphs
  • End-to-end automated refactoring pipeline with visual confirmation
  • Stable performance on medium-to-large repositories

What’s next

  • WebSocket-based live updates
  • Collaborative multi-user editing
  • Automated pull-request generation
  • Multi-language parsing engines
  • Direct GitHub and GitLab integrations
  • Smarter risk analysis and rollback previews

What it does

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for RepoSort

Share this project:

Updates