> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.nvidia.com/dsx-air/sdk/v1-0-0/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.nvidia.com/dsx-air/_mcp/server.

# v1.0.0

## Docs

- [Examples](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-examples.md)
- [Basic Simulation Operations](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-examples-simulations.md)
- [Starting and Stopping Simulations](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-examples-start-stop-simulations.md)
- [Import and Export Functionality](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-examples-import-export.md)
- [Automatic Out-of-Band Management (OOB)](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-examples-oob.md)
- [Reading a Simulation's History](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-examples-history.md)
- [Working with ZTP Scripts](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-examples-ztp-scripts.md)
- [Images](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-examples-images.md)
- [Nodes](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-examples-nodes.md)
- [Workers, Fleets and Certificates](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-examples-workers-fleets.md)
- [air_sdk](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-air-sdk.md)
- [air_sdk.const](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-air-sdk-const.md)
- [air_sdk.endpoints](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-air-sdk-endpoints.md)
- [air_sdk.endpoints.fleets](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-air-sdk-endpoints-fleets.md)
- [air_sdk.endpoints.history](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-air-sdk-endpoints-history.md)
- [air_sdk.endpoints.images](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-air-sdk-endpoints-images.md)
- [air_sdk.endpoints.interfaces](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-air-sdk-endpoints-interfaces.md)
- [air_sdk.endpoints.manifests](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-air-sdk-endpoints-manifests.md)
- [air_sdk.endpoints.marketplace_demo_tags](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-air-sdk-endpoints-marketplace-demo-tags.md)
- [air_sdk.endpoints.marketplace_demos](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-air-sdk-endpoints-marketplace-demos.md)
- [air_sdk.endpoints.node_instructions](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-air-sdk-endpoints-node-instructions.md)
- [air_sdk.endpoints.nodes](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-air-sdk-endpoints-nodes.md)
- [air_sdk.endpoints.organizations](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-air-sdk-endpoints-organizations.md)
- [air_sdk.endpoints.services](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-air-sdk-endpoints-services.md)
- [air_sdk.endpoints.simulations](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-air-sdk-endpoints-simulations.md)
- [air_sdk.endpoints.ssh_keys](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-air-sdk-endpoints-ssh-keys.md)
- [air_sdk.endpoints.systems](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-air-sdk-endpoints-systems.md)
- [air_sdk.endpoints.user_configs](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-air-sdk-endpoints-user-configs.md)
- [air_sdk.endpoints.workers](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-air-sdk-endpoints-workers.md)
- [air_sdk.endpoints.ztp_scripts](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-air-sdk-endpoints-ztp-scripts.md)
- [air_sdk.exceptions](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-air-sdk-exceptions.md)
- [air_sdk.types](https://docs.nvidia.com/dsx-air/sdk/v1-0-0-air-sdk-types.md)


> **Note:** This page contains both a page directory (above) and the landing page content (below). The page directory is generated for agent use and does not appear on the landing page.

> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.nvidia.com/dsx-air/sdk/v1-0-0/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.nvidia.com/dsx-air/_mcp/server.

# Air SDK

This project provides a Python SDK for interacting with the NVIDIA Air API.

## Prerequisite

The SDK requires Python 3.10. There are a number of ways to achieve this such as `poetry`, `uv` or `virtualenv`.
The following shows how to achieve this on a few popular OS's using `virtualenv`.

```
# Install Python 3.10
brew install python@3.10  # MacOS
apt-get install python3.10  # Ubuntu/Debian
sudo pacman -S python310  # Arch

# Install, create and activate a Python 3.10 virtual environment
python3.10 -m pip install virtualenv
python3.10 -m virtualenv venv
. venv/bin/activate
```

## Installation

The package is available on [PyPI](https://pypi.org/project/nv-air-sdk/).

```bash
pip install nv-air-sdk==1.0.0
```

## Authentication Options

1. NGC Device Login with `AirApi.with_device_login(email=<YOUR_EMAIL>, org_num=<YOUR_NGC_ORG_NUMBER>)`
2. NGC CLI Login with `AirApi.with_ngc_config()` (or just `AirApi()`)
3. Explicitly specify an NGC API Key (also known as a Scoped API Key, or "SAK") by instantiating `AirApi.with_api_key(api_key=<YOUR_NGC_API_KEY>)`

To use option 2 you will need to have the [NGC CLI](https://org.ngc.nvidia.com/setup/installers/cli) installed and to have configured it with `ngc config set`. This will create a file at `~/.ngc/config` which contains your NGC credentials that will be auto-detected by the Air SDK.

We do not support NGC service keys yet.

## Examples

[Examples overview](/sdk/v1-0-0-examples)

## API Reference

[air_sdk API reference](/sdk/v1-0-0-air-sdk)