Overview: https://github.com/zonca/custom-jupyterhub-docker-image is a template that ships a JupyterHub-ready single-user image with common scientific Python tooling and sensible defaults. Just edit requirements.txt in GitHub and the built-in GitHub Actions workflow auto-builds and publishes a new image—no local Docker required. See the README for the detailed package list and CI steps.
How to use it in JupyterHub:
- Build or pull the image, then point your spawner at it, e.g.
c.DockerSpawner.image = "ghcr.io/your-org/custom-jupyterhub-docker-image:TAG"(or the equivalent field for your spawner/Helm chart). - Restart JupyterHub so new servers launch with the custom image.
Adopt it via GitHub templates: click “Use this template” to create your own copy (no fork history), tweak requirements.txt, and push to your namespace to trigger the build.
Where the image lives: the workflow can publish to GitHub Container Registry. Publish/pull docs: https://docs.github.com/en/packages/working-with-a-github-packages-registry/working-with-the-container-registry.