Subsystem: Model & Provider Layer

The model subsystem (src/model/) interfaces with language model reasoning engines.

NVIDIA NIM Exclusivity

Production model invocation is handled exclusively by NvidiaProvider (src/model/provider/nvidia.rs), communicating via OpenAI-compatible Server-Sent Events (SSE) streaming with https://integrate.api.nvidia.com/v1.

Masked Secrets

API keys are wrapped in MaskedSecret (src/config/provider_registry.rs). The raw key is exposed only when constructing HTTP Authorization: Bearer ... headers. In all logs, events, and TUI projections, it is masked (nvap...[MASKED]).