FAQ

Which platforms are supported?#

The current published package @neuro.corp/dash bundles a native dash.exe, so it runs on Windows hosts with Node.js 20 or newer. Native builds for macOS and Linux are not part of this distribution yet.

Which AI providers are supported?#

NVIDIA, through the OpenAI-compatible NIM API (integrate.api.nvidia.com), is the only integrated provider. Your key and selected model are stored in ~/.dash/config.json. OpenAI, Anthropic, Google and other providers are not supported in this release.

Where do my credentials live?#

Only on your machine. The key pasted via /providers is encrypted (AES-256-GCM) before being written to ~/.dash/config.json, and is transmitted only to the NVIDIA NIM endpoint as a bearer token on model requests. There is no Neuro-hosted account system for Dash today.

How is Dash installed?#

Two commands:

npm install -g "@neuro.corp/dash" · then run: dash

Are free models included?#

Dash does not ship its own inference. Using Dash with model features requires your own NVIDIA API key, and NVIDIA’s terms and quota apply. Generating or composing files with no valid key is not possible — Dash will ask you to set one with /providers.

What are the known limitations?#

• Single provider today (NVIDIA NIM).
• No headless/non-interactive mode; the TUI is the only interface.
• Options like --help or --version are not implemented — launching dash opens the TUI directly.
• Model catalog depends entirely on what NVIDIA serves for your key.