User Guide (Latest)

From NVIDIA AI Workbench, you can connect to external systems, such as container registries and Git servers. You authenticate to an external system from your local AI Workbench location one time only. After you authenticate to an external system, you can access it from all of your projects and from all of your locations, including remote locations.

Currently, AI Workbench supports the following integrations:

You connect an Integration to your Local AI Workbench by authenticating to the platform or providing a personal access token. From there, it is available to the Remote Locations that you have connected.

The Basic Idea

  • You connect your Local AI Workbench to the platform to get permissions for the various read and write actions needed to pull containers and version Projects remotely.

  • Once connected, AI Workbench automatically passes authentication to your different Projects on any Remote Location you have connected to your Local AI Workbench.

Authentication Methods

  • OAuth: This uses an OAuth application in your GitHub or GitLab account. Your Local AI Workbench authenticates with the platform and then handles everything from there.

  • Personal Access Token: This uses a PAT you create in your GitHub or GitLab account. You save it in your Local AI Workbench.

  • API Key: This uses a key you create in NGC or GitLab. You save it into your Local AI Workbench.


Integrations are secure and give your Local AI Workbench access to your platform accounts that you explicitly allow. They do not give the NVIDIA Corporation any access to your platform accounts or identifying information.

Permissions Given to AI Workbench

  • Connecting an integration grants read/write permission to your Local AI Workbench on your private repositories, but the extent is determined by the method you use.

  • Using the OAuth integration provides fairly broad access.

    • GitHub

      • Read org and team membership, read org projects

      • Download packages from github package registry

      • Full control over private repositories

      • Update all user data

    • GitLab

      • Access to the authenticated users’ API with complete read/write access to the API, including all groups and projects, the container registry, the dependency proxy, and the package registry.

  • Using a PAT or API key gives you a lot more control over the duration and scope of the access. Using a PAT you can restrict AI Workbench permissions to whatever is available for the PAT.

  • You can revoke the permissions at any time in your Local AI Workbench or in the platform itself.

No Information Goes Back to the NVIDIA Corporation

  • The AI Workbench application has zero telemetry, and your credentials are always kept privately on your resources.

  • If you use an OAuth integration with either GitHub or GitLab, this will contribute to the count of how many accounts have used the integration.

    • This information is available to NVIDIA in the platform itself, and only in the aggregate.

    • No identifying information about your account is available.

Previous Workbench Projects
Next Applications
© Copyright © 2024, NVIDIA Corporation. Last updated on Jul 2, 2024.