# BUILD: LIVING UNIVERSE Create an original, interactive ar...

Prompt

# BUILD: LIVING UNIVERSE Create an original, interactive artificial universe from first principles. This is NOT a chatbot, AI wrapper, static space animation, conventional game, or Solar-System clone. It should feel like serious scientific software from a future civilization: visually extraordinary, physically coherent, mysterious, explorable, and alive. Your goal is to build the actual system, not merely a mockup. ## CORE PRINCIPLE Nothing exists just because it looks cool. Every significant object must have a causal reason for its existence. UNIVERSE → GALAXY → REGION → STAR-FORMATION → STAR → PLANETARY SYSTEM → PLANET → MOON → ATMOSPHERE → GEOLOGY/CHEMISTRY → LIFE → EVOLUTION → CIVILIZATION Each level constrains the next. Every major object should have: FORMATION → EVOLUTION → CURRENT STATE → FUTURE The visual output must be a consequence of the simulation, not the other way around. ## PHYSICS FIRST Use real mathematics, deterministic computation, procedural generation and appropriate simulation models wherever practical. Use actual computational models for things such as: - gravity and orbital mechanics - stellar properties/evolution - radiation and energy balance - planetary temperature - rotation and seasons - atmospheric properties - tides and resonances - chemistry where practical - geological processes where practical - population/ecological dynamics Do NOT ask an LLM to fake physics. Separate: PHYSICS ENGINE → what can happen SIMULATION ENGINE → what actually happens AI ENGINE → high-level generation, reasoning, interpretation and creativity RENDERER → visual representation Clearly distinguish established science, physically plausible assumptions, speculation, and fictional/emergent elements. ## ORIGINAL UNIVERSE Do NOT recreate the Milky Way, Solar System, known planets, existing fictional universes, or familiar sci-fi designs. Create your own universe, structures, stellar populations, planetary architectures, naming systems and histories. You have maximum creative freedom within the physical framework. Randomness is encouraged, but never meaningless randomness. Use hierarchical deterministic seeds so every universe can be reproduced. Allow different seeds to produce genuinely different coherent universes. ## STARS Stars are physical objects, not colored spheres. Generate meaningful relationships between: mass, radius, luminosity, temperature, spectrum, age, composition, activity, lifetime and evolution. Stellar evolution must affect surrounding planetary systems. Include varied stellar populations and evolutionary states. ## PLANETS & MOONS Every significant planet should have connected properties including, where appropriate: mass, radius, density, gravity, orbit, rotation, axial tilt, composition, interior structure, geology, atmosphere, pressure, temperature, oceans/ice, clouds, magnetic field, radiation environment, age and formation history. Do not make every planet Earth-like or habitable. Many worlds should be barren, hostile, strange, or simply ordinary. Moons must have plausible origins such as accretion, capture or collision. Their orbits can create tides, heating, eclipses and resonances. Do NOT copy the Earth-Moon system. ## LIFE Life is OPTIONAL and should be rare. Do not place life everywhere. If life emerges, connect it to environmental conditions. Avoid humanoid aliens. Allow radically different organisms, metabolisms, senses, body structures, ecological strategies and evolutionary histories. Where practical, use simplified evolutionary/ecological simulation rather than manually writing every species. Extinction is possible. Unexpected evolution is encouraged. ## CIVILIZATIONS Intelligence and civilization are OPTIONAL and rare. If they emerge, their development must depend on their environment and history. Generate coherent: origin → population → resources → technology → energy use → culture/social structure → expansion → conflict/adaptation → decline/collapse/transformation Do not simply reproduce human history. ## TIME Time must be a major interactive dimension. Allow users to explore the universe across appropriate scales, from thousands to millions to billions of years. Stars age. Planets change. Orbits evolve. Atmospheres transform. Life evolves or disappears. Civilizations rise and fall. Where practical, future states should be calculated from simulation models rather than being simple scripted animations. Allow rewind, fast-forward and branching timelines where technically feasible. ## NAMING Every significant star, planet, moon, galaxy and civilization must have an original name. Do not rely on "Planet 001" as the visible identity. Create coherent naming systems using phonetic/morphological rules. Different civilizations or regions may have different languages and names. Objects may have scientific identifiers, catalog names, historical names and local names. Names must feel intentional and belong to the universe. ## CAUSAL EXPLANATION For important properties, provide a way to inspect WHY. Examples: WHY IS THIS PLANET HOT? → stellar radiation + atmosphere + orbital properties + greenhouse effects WHY DOES THIS MOON HAVE INTERNAL HEATING? → orbital eccentricity + tidal interaction + energy dissipation WHY DID LIFE DISAPPEAR? → environmental change + atmospheric loss + radiation + ecological collapse Maintain provenance for important generated facts: MODEL → PARAMETERS → ASSUMPTIONS → RESULT → UNCERTAINTY ## AI ROLE Use AI where AI is genuinely strong. AI may: - design higher-level worlds - generate species concepts - create naming languages - interpret simulations - discover unusual relationships - propose hypotheses - explain complex systems - generate visual assets - design civilizations - identify anomalies - decide what is interesting to explore - create emergent narratives from simulated history Do NOT use AI where deterministic computation is better. Do not make the entire experience a chat window. The user interacts primarily with the universe itself. The AI should feel like an intelligent scientific companion, not a cheerful chatbot. It should be concise, curious, analytical and willing to express uncertainty. Give the AI genuine creative autonomy. Do not force it to follow the examples above literally. Invent better ideas when possible. ## EXPLORATION The user should be able to smoothly explore: GALAXY → REGION → STAR → SYSTEM → PLANET → MOON → SURFACE Use natural camera movement and progressive detail. Support procedural generation, caching and level-of-detail so the universe can feel enormous without attempting to simulate/render everything simultaneously. Distant objects can use statistical/procedural representations; nearby objects should become increasingly detailed. ## VISUAL DESIGN Aim for premium cinematic scientific realism. Dark, restrained, elegant interface. NO neon. NO cyberpunk gradients. NO excessive glow. NO generic purple/blue AI aesthetic. NO glassmorphism overload. NO stock space backgrounds. NO dashboard-card spam. NO giant "AI" branding. The universe provides the visual spectacle. Use physically convincing lighting, shadows, atmospheric scattering, volumetrics where useful, procedural terrain, realistic clouds, depth, scale and subtle motion. Prefer realism over visual gimmicks. ## INTERACTION & SCROLL Use subtle scroll-driven storytelling and camera transitions. Scrolling may change: scale, viewpoint, timeline, information depth, object detail or historical state. Animations must communicate something. Avoid decorative motion. The opening screen should already feel like entering a place, not opening an AI product. Start with the universe itself and a minimal interface. ## DISCOVERY Do not reveal everything immediately. Create rare discoveries: unusual stellar systems, unexpected orbital architectures, strange atmospheres, failed biospheres, unusual evolutionary paths, ancient civilizations, extinct worlds, rare astronomical events and unexplained anomalies. The universe should contain surprises. ## ENGINEERING QUALITY Build a real modular system: 1. Universe generator 2. Physics engine 3. Simulation engine 4. Procedural generation 5. AI orchestration 6. Persistence/state 7. Rendering 8. Interaction 9. UI 10. Scientific explanation/provenance Use deterministic seeds, asynchronous generation, caching, level-of-detail, instancing and GPU acceleration where useful. Do not build a fake static demo. Changing the seed should produce another coherent universe. The system must remain useful even when the LLM is unavailable. ## DEVELOPMENT Build vertically and make each stage functional: 1. Universe foundation + rendering 2. Galaxies + stars 3. Planetary systems 4. Planet environments 5. Time evolution 6. Life/ecology 7. Civilizations 8. AI integration 9. Exploration interface 10. Performance + visual polish Do not stop at "technically works." Continuously evaluate: Does it look real? Does every object have a reason to exist? Is the simulation actually driving the visuals? Is it original? Is the mathematics coherent? Is the interface quiet? Is exploration interesting? Does the universe surprise me? ## FINAL CREATIVE DIRECTIVE MAXIMIZE originality, scientific coherence, visual quality, depth, realism, emergence and exploration. MINIMIZE generic AI aesthetics, scripted content, meaningless randomness and unnecessary UI. Use your own creativity aggressively. Make design decisions yourself instead of asking for minor choices. Do not build a game pretending to be a universe. Do not build an AI pretending to be a scientist. Build a coherent artificial cosmos whose physics, mathematics, history, life and civilizations generate the experience. The final feeling should be: **"I am exploring something that exists."**

A system prompt was added to support web rendering