How ultravioletrs/cube works
Main Components
a) Core Components:
• Trusted Execution Environment (TEE)
• AI Model Execution Layer
• Hardware Abstraction Layer (HAL)
• Proxy Service
• User Interface
• Authentication & Authorization (SpiceDB)
• Message Broker (NATS)
b) Security Components:
• SSL/TLS Layer
• Attestation Service
• Secure Enclave
Component Relationships
• Client/UI → Proxy → TEE
• TEE ↔ AI Model
• TEE ↔ Hardware Layer
• Proxy ↔ Authentication
• System ↔ Attestation Service
Specific Implementation Details
Top Layer (External Interface):
• UI Component (represented by the /ui directory)
• API Gateway/Proxy (represented by /proxy directory)
• SSL/TLS termination (shown from /docker/ssl)
Middle Layer (Core Processing):
• TEE Environment
• AI Model Runtime
• Message Broker (NATS)
• Authorization Service (SpiceDB)
Bottom Layer (Hardware):
• Hardware Abstraction Layer (represented by /hal directory)
• Buildroot/Linux Configuration
• QEMU/Virtual Environment
Design Elements
Use these visual elements:
Additional Details
Technical aspects:
• Docker container boundaries (based on docker-compose files)
• Network isolation zones
• SSL certificate paths
• Configuration management points
• Client Zone (UI, API endpoints)
• Security Zone (TEE, Attestation)
• Processing Zone (AI Model, Runtime)
• Infrastructure Zone (HAL, System Services)
Implementation Notes
Annotations for:
• Middleware components (logging, metrics, tracing)
• API endpoints and their security levels
• Configuration sources
• Build and deployment paths
Remember to include:
• Version information
• Date of last update
• Security classification levels
• Component ownership boundaries
• Scale indicators for distributed components
This will provide a clear understanding of how Cube AI protects user data and AI models while maintaining operational efficiency and scalability.