I designed Satori AI’s command architecture as a strict JSON‑only interface that guarantees deterministic, auditable scene changes and powers the undo/redo system.
Scaling factor for the "Make it 20% larger" prompt is 1.2.
Local development backend runs on port 8000.
CORS whitelist for development is "http://localhost:5173,http://localhost:3000".
Gemini model used by the backend is gemini-2.5-flash.
For more context on the core features of Satori AI, see satori ai core features.