Kalpa ai

see preview and do vibe coding

Comment

Inspiration

  • Developers constantly switch between tools (editor → terminal → browser → devices → AI assistants).
  • This breaks focus and slows productivity.
  • We wanted one unified browser-based development environment powered by AI.

What it does

  • Full web IDE with Monaco editor, terminal, file system, and PWA support.
  • Supports multiple projects open at the same time in one editor.
  • AI-powered coding with 5 AI providers (OpenAI, Anthropic, Gemini, Groq, Ollama).
  • Generates complete project plans, designs, tasks, and assets.
  • Real device integration for Android & iOS (logs, commands, apps).
  • Cloud integration with Supabase, Firebase, GitHub.
  • Organization-level integrations: GitHub, Bitbucket, Jira for workflow automation.
  • Offline support with service workers & IndexedDB.

How we built it

  • Spec-driven architecture: requirements → design → tasks.
  • AI-assisted development (~90% code generated).
  • Frontend: React, TypeScript, Vite, Monaco.
  • Backend: Node.js, Express, WebSocket, node-pty.
  • Automated testing, linting, and type-check pipelines.

Challenges we ran into

  • Making Monaco Editor work smoothly inside React.
  • Managing multiple WebSocket connections across devices & terminals.
  • Unifying Android (ADB) & iOS (libimobiledevice) APIs.
  • AI provider rate limits requiring automatic fallback.
  • Enabling multi-project handling inside one workspace.
  • Adding GitHub, Bitbucket & Jira authentication + organization permissions.

Accomplishments that we're proud of

  • Built a production-ready IDE in 3 weeks (instead of 10).
  • 72,000+ lines of code, 150+ components, 25+ backend services.
  • Real device testing directly from the browser.
  • Multi-AI provider fallback → 99.9% uptime.
  • Multi-project workspace support.
  • Seamless GitHub, Bitbucket & Jira integration for enterprise users.

What we learned

  • Strong specifications + AI = huge productivity boost.
  • AI handles patterns well but architects must define the system.
  • Property-based testing uncovers rare bugs early.
  • Modern browser APIs make powerful developer tools possible.
  • TypeScript prevents large-scale errors, especially with AI-generated code.
  • Organization integrations require strict role/permission handling.

What's next for Kalpa AI

  • Real-time collaboration (multi-cursor editing, shared terminal).
  • Extension marketplace for themes, languages, and tools.
  • Mobile app for coding anywhere.
  • Docker & Kubernetes container environments.
  • Visual CI/CD builder with drag-and-drop workflows.
  • Enterprise edition with self-hosting, SSO, custom AI models.
  • Deeper DevOps integrations (GitHub Actions, Bitbucket Pipelines, Jira automation).

Built With

Share this project:

Updates