@royalrat-pack/pi-extension-nvidia-build-provider

A Pi extension that adds NVIDIA Build API models as a first-class provider with native /login support and /model selection.

What this extension does

• Registers provider: nvidia-build
• Adds login option in Pi: NVIDIA Build API
• Stores API key in Pi auth storage after one-time paste
• Enables model selection through /model
• Supports model auto-discovery via NVIDIA /v1/models

Installation

Option 1: Install from npm (recommended)

pi install npm:@royalrat-pack/pi-extension-nvidia-build-provider

Option 2: Install from GitHub

pi install git:github.com/debarch777/pi-extension-nvidia-build-provider

Option 3: Local development install

pi install /absolute/path/to/pi-extension-nvidia-build-provider

Usage

Inside Pi:

/login
# select: NVIDIA Build API
# paste API key once

/model
# select nvidia-build/<model-id>

Optional environment variables

• NVIDIA_BASE_URL (default: https://integrate.api.nvidia.com/v1)
• NVIDIA_MODELS (comma-separated model IDs; skips auto-discovery)
• NVIDIA_INCLUDE_ALL_MODELS=1 (includes non-chat model IDs too)
• NVIDIA_API_KEY or NVCF_API_KEY (optional startup fallback)

Package

• npm: @royalrat-pack/pi-extension-nvidia-build-provider
• latest published: 0.1.0

Development

git clone https://github.com/debarch777/pi-extension-nvidia-build-provider.git
cd pi-extension-nvidia-build-provider
pi -e .

Security note

Only install Pi extensions from sources you trust. Extensions execute with your user permissions.

License

MIT