name	project-scaffold
description	Bootstrap new biomedical ML research repositories with production-grade structure. Use when: (1) Creating a new research project from scratch, (2) Setting up pyproject.toml with uv, (3) Establishing directory layout, CLAUDE.md, or config patterns for a biomedical/healthcare ML project.

Project Scaffold

Quick Start

python scripts/init_project.py my-study --path /path/to/workspace --python-version 3.11
python scripts/init_project.py my-study --path /path/to/workspace --python-version 3.11 --docker --dependency monai
cd my-study && uv sync && uv run pytest tests/

Creates a uv-managed Python project with: src/<pkg>/config.py for paths/settings, tests/, a dated docs/<YYYYMMDD>/ directory, CLAUDE.md, pre-commit hooks, ruff config, and standard biomedical ML dependencies.

What Gets Created

my-study/
├── pyproject.toml          # uv/hatch config + dependencies + ruff config
├── .python-version         # Pinned Python version
├── .pre-commit-config.yaml # ruff linting + formatting
├── .env.example            # Template for secrets
├── .gitignore
├── CLAUDE.md               # Agent guidance
├── README.md
├── Dockerfile              # Optional, if --docker is passed
├── .dockerignore           # Optional, if --docker is passed
├── src/<pkg>/
│   ├── __init__.py
│   └── config.py           # Paths and project settings (env var overrides)
├── scripts/                # CLI entry points
├── tests/
│   └── test_config.py
└── docs/<YYYYMMDD>/        # Dated reports and figures

Decision Points

ASK the user before scaffolding:

What Python version (3.10, 3.11, 3.12)?
Do they need Docker support for reproducible environments?
Any additional dependencies beyond the defaults (numpy, pandas, scikit-learn, scipy, tqdm, matplotlib, wandb)?

References

File	Read When
references/project-structure.md	Directory layout conventions, uv setup, config patterns, CLAUDE.md template, pre-commit/ruff setup, Docker support, env var management

Scripts

Script	Purpose
`scripts/init_project.py`	Scaffold a minimal new project (supports `--python-version`, `--docker`, and repeatable `--dependency` flags)

Project Scaffold

Quick Start

python scripts/init_project.py my-study --path /path/to/workspace --python-version 3.11 python scripts/init_project.py my-study --path /path/to/workspace --python-version 3.11 --docker --dependency monai cd my-study && uv sync && uv run pytest tests/

What Gets Created

my-study/ ├── pyproject.toml # uv/hatch config + dependencies + ruff config ├── .python-version # Pinned Python version ├── .pre-commit-config.yaml # ruff linting + formatting ├── .env.example # Template for secrets ├── .gitignore ├── CLAUDE.md # Agent guidance ├── README.md ├── Dockerfile # Optional, if --docker is passed ├── .dockerignore # Optional, if --docker is passed ├── src/<pkg>/ │ ├── __init__.py │ └── config.py # Paths and project settings (env var overrides) ├── scripts/ # CLI entry points ├── tests/ │ └── test_config.py └── docs/<YYYYMMDD>/ # Dated reports and figures

Decision Points

ASK the user before scaffolding:

What Python version (3.10, 3.11, 3.12)?

Do they need Docker support for reproducible environments?

Any additional dependencies beyond the defaults (numpy, pandas, scikit-learn, scipy, tqdm, matplotlib, wandb)?

References

File

Read When

references/project-structure.md

Directory layout conventions, uv setup, config patterns, CLAUDE.md template, pre-commit/ruff setup, Docker support, env var management

Scripts

Script

Purpose

scripts/init_project.py

Scaffold a minimal new project (supports --python-version, --docker, and repeatable --dependency flags)

project-scaffold