dash-init

Project generator for Plotly Dash: one command to a running app.

pip install dash-init
dash-init create my-dashboard
cd my-dashboard
pip install -r requirements.txt     # or: uv sync
python app.py                       # or: uv run app.py

You get a working, current Dash app with a callback, a chart, an assets/ folder, a README, requirements.txt + pyproject.toml, and a .gitignore: the structure you'd have written yourself.

my-dashboard/
├── app.py              # layout + callback, exposes server for deployment
├── assets/
│   └── custom.css      # custom styling
├── requirements.txt    # dependencies
├── pyproject.toml      # project metadata and dependencies
├── README.md           # how to run and deploy it
└── .gitignore

Every generated project includes a README with steps to deploy it (Plotly Cloud, any Python host, or Docker with --docker).

Generated projects target Python 3.11+ and pin current versions: dash>=4.4.1, pandas>=3.0.5, dash-bootstrap-components>=2.0.4 (bootstrap template). The dash-init tool itself runs on Python 3.9+ and has no dependencies.

Usage

dash-init create <name>                      create a project from the default (minimal) template
dash-init create .                           scaffold into the current directory (see below)
dash-init create <name> -t <template>        choose a built-in or hub template
dash-init create <name> -t gh:owner/repo/sub/dir[@ref]
                                            pull any directory on GitHub as a template
dash-init create <name> --docker             also generate a Dockerfile + .dockerignore (alias: --dockerfile)
dash-init templates                          list everything available
dash-init --version

Into an existing project

dash-init create . scaffolds into the current directory instead of making a new one. It never overwrites a file you already have, and it plays nicely with uv:

mkdir my-dashboard && cd my-dashboard
uv init
uvx dash-init create .

If a pyproject.toml is already present, dash-init leaves it alone and runs uv add "dash>=4.4.1" ... for you, so your file keeps its formatting and the dependencies land in uv.lock and your venv. Pass --no-install to print the command instead of running it. Without uv on your PATH, the commands to run are printed either way.

create <name> refuses to overwrite a non-empty directory and never installs anything: it writes files and prints the next steps. Only create . with an existing pyproject.toml runs uv add (announced first, skippable with --no-install).

Built-in templates

These ship inside the package and need no network. Every one is generated and smoke-run in CI against current Dash.

Add --docker to any of them for a ready-to-build Dockerfile + .dockerignore (docker build -t my-dashboard . && docker run -p 8050:8050 my-dashboard).

Templates from dash-templates-hub

Full admin dashboards from budescode/dash-templates-hub are available by name and downloaded when you run create:

dash-init create school -t kiaalap

What happens:

• Only that template's files are fetched: no git clone, no full-repo tarball. Repo clutter (screenshots, notebooks, caches, lockfiles) is dropped.
• The project name is set in pyproject.toml; a pyproject.toml (from requirements.txt) and a .gitignore are generated if the template lacks them.
• The hub is pinned to a specific commit in src/dash_init/registry.json, and CI fetches and checks every entry, so a change in the hub cannot silently break create.
• A failed download leaves nothing behind. Errors say whether you're offline, the path doesn't exist, or you hit GitHub's rate limit (set GITHUB_TOKEN to raise it).

Any GitHub directory works the same way, un-curated:

dash-init create demo -t gh:someone/repo/examples/my-app@v2.1

Why trust it

• Tested against current Dash. Every template is generated and booted in CI, weekly, so if Dash breaks something you see a red build here, not in your terminal.
• Never touches your environment. It writes files and tells you what to run. No pip install on your behalf.
• Zero dependencies. Pure standard library, so installing it can't conflict with anything.

Roadmap

• deploy subcommand
• --dir option to create the project outside the current directory
• Turn an existing app into a template (dash-init create --from-project)
• More templates

Contributing

Contributions are welcome. See CONTRIBUTING.md for guidelines.

• Report a bug
• Request a feature

Connect

• PyPI: https://pypi.org/project/dash-init
• LinkedIn: https://www.linkedin.com/in/budescode
• PayPal: https://www.paypal.com/paypalme/omonbudeemma

Prior art & credit

The template-directory approach follows the archived dash-tools by Andrew Hossack (MIT). No code was reused, but the idea that a template should be a plain folder anyone can contribute to comes from there.

License

MIT