GitHub-specific scopes#

The following GitHub scopes may be suitable for certain use cases:

read:org grants access to the users’ organizations. This is handy if you want to use GitHub organizations in your backend environment as Unix groups for collaboration purposes. Having globally consistent UIDs (from the GitHub ID) and GIDs (from the organization IDs) makes access permissions on shared storage much easier.

public_repo allows “trusted users” read and write privileges for public repositories. If you want to automatically provision git pushes to GitHub, you can accomplish this by passing a token with this scope to your Lab or classic Notebook instance.

repo does the same for private repositories too.

user:email allows the authenticator to determine email addresses even if they are marked private. Having access to email addresses, in conjunction with read/write repository access, allows preconfiguring the user’s git configuration for GitHub pushes without any required action by the user.

The additional fields exposed by expanded scope are all stored in the authenticator’s auth_state structure, so you’ll need to enable auth_state and install the Python cryptography package to be able to use these.

We currently use the following fields:

  • id is an integer set to the GitHub account ID.

  • login is the GitHub username

  • name is the full name GitHub knows the user by.

  • email is the publicly visible email address (if any) for the user.

  • access_token is the token used to authenticate to GitHub.

  • teams is list of teams the user is part of, fetched only if populate_teams_in_auth_state option is set to True on GitHubOAuthenticator. The read:org scope is also required for this to work.

To use this expanded user information, you will need to subclass your current spawner and modify the subclass to read these fields from auth_state and then use this information to provision your Notebook or Lab user.