Vision: A real-time collaborative notebook system enabling seamless AI ↔ Python ↔ User interactions through local-first architecture.
Current Status: Production system deployed at https://app.runt.run with real-time collaboration, AI integration, and Python execution. Focus is on alpha release preparation with improved runtime management.
- Real-time collaborative notebooks at https://app.runt.run
- Python execution via Pyodide with rich outputs (matplotlib SVG, pandas HTML, IPython.display)
- AI integration with full notebook context awareness and tool calling
- Event-sourced architecture with LiveStore providing reliable state management
- Artifact service for large outputs with R2 storage
- Authentication system with OIDC OAuth
- Mobile support with responsive design and keyboard optimizations
- Offline-first operation that syncs when connected
Users can create notebooks, execute Python code, collaborate in real-time, use AI assistants that see code and outputs, and work offline. The system is stable and handles production workloads.
Goal: Eliminate manual runtime setup friction for alpha users
Current State: Users must manually run NOTEBOOK_ID=xyz pnpm dev:runtime
Target: One-click runtime startup directly in browser
Implementation (#447):
- Browser-based Pyodide runtime agent
- Embedded runtime management in web client
- Auto-connect when opening notebooks
- Owner-only runtime launch permissions
Benefits:
- Zero-friction notebook experience
- Scalable for alpha user onboarding
- No infrastructure management needed
Stale Runtime Cleanup (#465):
- Button to evict disappeared runtimes
- Runtime health monitoring
- Automatic cleanup of stale connections
RuntHQ Integration (#335):
- Autoconnect runtime on notebook open
- Seamless runtime orchestration
- Owner permission controls
Finalize Branding (#448):
- Complete transition to "In the Loop" branding
- Domain strategy for alpha launch
LiveStore Output Reactivity (#415 - Critical):
- Fix UI becoming unresponsive when outputs update
- React Compiler compatibility issues
- Ensure reliable real-time collaboration
Local CPython Support (#87, #99):
- Connect to external Python environments
- Ephemeral and persistent environment options
Improved Code Editing (#154):
- Tab completion (Jupyter-style)
- Better autocomplete behavior
- LSP integration foundations
Output Management (#173):
- Granular output collapsing
Multi-Runtime Support:
- ZeroMQ integration for external kernels (#61)
- Jupyter kernel protocol compatibility
- Container and remote runtime management
Enhanced Python Experience:
- Package management system (#63)
- Variable inspection and debugging (#175)
- SQL cell implementation with DuckDB (#62)
Enhanced Outputs:
- Vega/Vega-Lite visualization (#267)
- GeoJSON mapping support (#266)
- Interactive widgets (#183)
- Advanced 3D visualizations
Content Management:
- Jupyter notebook compatibility (#184)
- Import/export capabilities
- Runtime startup: < 10 seconds (one-click browser launch)
- User onboarding: < 2 minutes from signup to first execution
- System reliability: > 99% uptime for production notebooks
- Collaboration latency: < 100ms for real-time updates
- Test suite: < 30 seconds execution time
- Build time: < 5 minutes for full deployment
- Memory efficiency: Support 100+ cell notebooks
- Cross-browser support: Chrome, Firefox, Safari compatibility