Build Prompt: Qwen Agent Bot in Three.js
Create a complete, self-contained index.html that renders an impressive 3D Qwen Agent Bot using Three.js.
The bot is the physical embodiment of a powerful LLM agent. It should look sleek, intelligent, futuristic, and believable enough to exist as a real machine.
Visual Design
Build a humanoid sci-fi robot with:
a smooth angular head and glowing visor
a transparent or partially open chest containing a luminous AI core
dark graphite mechanical structure with white ceramic armor panels
violet, blue, and cyan emissive lighting
articulated arms, hands, legs, shoulders, and neck
visible cooling vents, cables, joints, and layered mechanical parts
a strong, readable silhouette
premium cinematic hard-surface design
Add several glowing Mixture-of-Experts modules around the chest, shoulders, or upper back. Some modules should pulse brightly while others remain dim. Animate light traveling between the head, chest core, expert modules, and arms to represent token routing and active reasoning.
Give the bot a confident standing pose with one hand interacting with a floating holographic interface.
Three.js Requirements
Produce one working HTML file containing all CSS and JavaScript.
Use:
Three.js from a CDN
OrbitControls
procedural geometry made from Three.js primitives and custom mesh combinations
physically based materials
emissive materials
shadows
cinematic lighting
a dark futuristic environment
subtle fog or particles
a reflective or metallic platform
responsive full-screen rendering
Add animation for:
rotating chest-core rings
pulsing expert modules
moving data lights
subtle head movement
holographic UI motion
gentle idle mechanical motion
The holographic interface should show abstract agent activity, token streams, expert activation, and system status using simple glowing shapes, lines, and panels.
Interaction
Include:
mouse orbit
zoom
smooth camera controls
automatic slow camera rotation when idle
a small UI button to pause or resume animation
a small UI button to reset the camera
Output
Return only the complete contents of index.html.
The result should run immediately when opened in a browser and should look polished enough to serve as a visual prototype for a real-time LLM monitoring interface.