https://github.com/nebari-dev/nebari-environments
Community-contributed Pixi environments for data science, ML, and scientific computing — published as OCI artifacts via nebi
https://github.com/nebari-dev/nebari-environments
Last synced: 16 days ago
JSON representation
Community-contributed Pixi environments for data science, ML, and scientific computing — published as OCI artifacts via nebi
- Host: GitHub
- URL: https://github.com/nebari-dev/nebari-environments
- Owner: nebari-dev
- License: apache-2.0
- Created: 2026-02-13T18:41:59.000Z (5 months ago)
- Default Branch: main
- Last Pushed: 2026-02-17T22:39:56.000Z (4 months ago)
- Last Synced: 2026-03-11T13:33:45.737Z (4 months ago)
- Homepage: https://quay.io/organization/nebari_environments
- Size: 244 KB
- Stars: 1
- Watchers: 0
- Forks: 1
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Nebari Environments
[](https://github.com/nebari-dev/nebari-environments/actions/workflows/publish.yml)
A collection of community-contributed, ready-to-use [Pixi](https://pixi.sh) environments for data science, machine learning, and scientific computing. Environments are automatically published as OCI artifacts to [`quay.io/nebari_environments`](https://quay.io/organization/nebari_environments) and can be imported with a single command using the [nebi](https://github.com/nebari-dev/nebi) CLI.
## Using an environment
Install the [nebi](https://github.com/nebari-dev/nebi) CLI, then import any published environment:
```bash
nebi import quay.io/nebari_environments/data-science-demo:v1
```
This pulls the `pixi.toml` and `pixi.lock` into the current directory. Then install and activate with [Pixi](https://pixi.sh):
```bash
pixi install
pixi shell
```
You can also import into a specific directory or as a global workspace:
```bash
# Import into a specific directory
nebi import quay.io/nebari_environments/data-science-demo:v1 -o ./my-project
# Import as a global workspace
nebi import quay.io/nebari_environments/data-science-demo:v1 --global data-science
```
Browse available environments at [`quay.io/nebari_environments`](https://quay.io/organization/nebari_environments).
## Contributing an environment
1. Create a new directory under `environments/`:
```
mkdir environments/my-env
```
2. Add a `pixi.toml` with a `[workspace]` section:
```toml
[workspace]
name = "my-env"
channels = ["conda-forge"]
platforms = ["linux-64", "linux-aarch64", "osx-arm64", "osx-64"]
version = "0.1.0"
[dependencies]
python = ">=3.11"
# add your packages here
```
3. Open a pull request. Once merged to `main`, the environment will be automatically published to `quay.io/nebari_environments/my-env`.
## Repository structure
```
environments/
/
pixi.toml
```
Each directory under `environments/` is a standalone Pixi environment. The directory name is used as the environment name when publishing.
## CI / Publishing
A GitHub Actions workflow (`.github/workflows/publish.yml`) runs on every push to `main`. It:
1. Downloads the [nebi](https://github.com/nebari-dev/nebi) CLI
2. Starts an ephemeral nebi server
3. Pushes and publishes every environment to `quay.io/nebari_environments/`
Nebi handles deduplication server-side, so all environments are published on every run.
### Required GitHub secrets
| Secret | Purpose |
|--------|---------|
| `QUAY_USERNAME` | quay.io robot account username |
| `QUAY_PASSWORD` | quay.io robot account password/token |
| `QUAY_API_TOKEN` | quay.io OAuth token for setting repos to public |
### Quay.io setup
The target organization is [`nebari_environments`](https://quay.io/organization/nebari_environments) on quay.io.
The robot account used by CI needs permission to **create new repositories** in the organization. By default, robot accounts cannot create repos. To grant this:
1. Go to the [nebari_environments org](https://quay.io/organization/nebari_environments) on quay.io
2. Navigate to **Teams and Membership**
3. Create a team (e.g. `ci-publishers`) or use an existing one
4. Set the team's **role to "Creator"** — this allows members to create new repositories
5. Add the robot account to that team (e.g. `nebari_environments+ci`)
The robot account credentials are then stored as `QUAY_USERNAME` and `QUAY_PASSWORD` GitHub secrets on this repository.
New repositories on quay.io are **private by default**. The workflow automatically sets each repository to public after publishing using the quay.io API. This requires an OAuth token (`QUAY_API_TOKEN`) with the **"Administer Repositories"** scope. To generate it:
1. In the same org, navigate to **Applications** (under org settings)
2. Create a new OAuth application (e.g. `ci-visibility`)
3. Generate a token with the **"Administer Repositories"** permission
4. Store the token as the `QUAY_API_TOKEN` GitHub secret on this repository