AI Workbench Integrations
Integrations are external systems and services, such as container registries and Git servers, that you connect to from NVIDIA AI Workbench to enable or expand functionality. After you connect, the integration is available in AI Workbench. For example, if you connect to Github, you can commit changes and publish your projects to your Github account from inside AI Workbench.
Some integrations are available in AI Workbench by default, and some custom integrations you can add to AI Workbench yourself as explained later. You must configure both default and custom integrations before you can use them. Currently, AI Workbench supports the following integrations.
Integration |
Information |
Default / Custom |
|---|---|---|
| GitHub | GitHub Integration | Default |
| GitLab | GitLab Integration | Default |
| Self-hosted GitLab | Self-Hosted Gitlab | Custom |
| NGC Private Registry | NGC Private Registry Integration | Default |
Use this documentation to learn about the following:
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.
Depending on the integration, you can authenticate by using one of the following methods:
OAuth — This method uses an OAuth application. AI Workbench authenticates with the platform and then handles everything from there. For more information, see NVIDIA AI Workbench Data Privacy and Data Collection.
Personal Access Token (PAT) — A PAT is a key that you create in your GitHub or GitLab account. You save it in AI Workbench. By using a PAT, you can restrict AI Workbench permissions to whatever is available for the PAT.
API Key — An API key is a key that you create in NGC. You save it into your Local AI Workbench.
Integrations do not give the NVIDIA Corporation any access to your platform accounts or identifying information. Your credentials are always kept privately on your resources. You can revoke permissions at any time in your Local AI Workbench or in the platform itself. For details, see NVIDIA AI Workbench Data Privacy and Data Collection.
When you disconnect an integration, AI Workbench deletes the credentials AI Workbench has stored for the connection.
You can view and modify your integrations by using the Workbench Settings window of the AI Workbench desktop app, or by using the CLI. The primary actions you perform with integrations are to connect to or disconnect from them.
The following is the Workbench Settings window.
Desktop App
Use the following procedure to connect to your Git account, or other integrations, by using the AI Workbench desktop app.
At the top of the AI Workbench window, click the gear icon.
The Workbench Settings window appears.
In the Workbench Settings window, select the Integrations page.
Click Connect or Disconnect on the tile of the integration that you want to modify.
CLI
Use the following commands to work with your integrations by using the CLI.
To get information about your integrations, run the following code.
nvwb list integrations
To connect to an integration, run the following code.
nvwb connect integration <integration name>
To disconnect from an integration, run the following code.
nvwb disconnect integration <integration name>
Currently, AI Workbench supports custom integrations with self-hosted GitLab servers.
Desktop App
Use the following procedure to add a custom integration by using the AI Workbench desktop app.
At the top of the AI Workbench window, click the gear icon.
The Workbench Settings window appears.
In the Workbench Settings window, select the Integrations page.
Click Add Integration.
The Add Integration window appears.
Specify Integration Type, Display Name, and Authentication Method.
For Integration Type, only Self-hosted GitLab is supported. Specify all additional information as explained in Self-Hosted Gitlab.
Click Add.
Your new integration appears on the Integrations page.
Click Connect on your new integration when you are ready to start using it.
CLI
Use the following procedure to add a custom integration by using the CLI.
To create an integration, run the following code. The interactive dialog starts.
nvwb create integration
Select
gitlaband press enter.Specify all additional information as explained in Self-Hosted Gitlab.
After you complete the dialog, the list of your integrations appears. Verify that you see your new custom integration in the list.
To connect to your new integration, run the following code.
nvwb connect integration <integration name>
You can use either an OAuth flow or a Personal Access Token (PAT) to authenticate with GitHub. When you create a PAT, assign the following scopes:
reporead:orguserread:packages
AI Workbench needs access to your private repositories on GitHub. Based on your actions, AI Workbench acts on your behalf to create public or private repositories, and to push changes to your public and private repositories.
For more information about authenticating to GitHub, see Authorizing OAuth applications in GitHub.
You can use either an OAuth flow or a Personal Access Token (PAT) to authenticate with Gitlab. When you create a PAT, assign the following scopes:
api
AI Workbench needs full API access to GitLab. Based on your actions, AI Workbench acts on your behalf to create public or private repositories, and to push changes to your public and private repositories. With Gitlab PATs, there is no other option that allows write access to create new repositories on your behalf.
AI Workbench integrates with NVIDIA’s NGC Catalog to access base environment images stored in the private registry. Use this integration only if you are part of a program that needs access to the NGC Private Registry.
Integration with the NGC Private Registry supports the API Key authentication method only. Do not use personal keys. To generate your API key, log into NVIDIA’s NGC Catalog, navigate to the Setup page, and then click Generate API Key.