Example project layouts

Use only the folders you need. See page 4 for Git and data organisation.

Start small

project/
├── README.md              # question and run command
├── CLAUDE.md              # agent instructions
├── pyproject.toml
├── uv.lock
├── .gitignore
├── data/
│   └── generated/         # fictional practice data
├── scripts/
│   └── analyse.py
├── results/
└── tests/

Ask the agent to create this structure in your practice project. Copy the text inside the instruction template, rather than its explanatory wrapper. The CSV for the first exercise contains fictional observations.

Add structure as the project grows

project/
├── MODEL.md               # scientist-checked equations and conventions
├── TASKS.md               # current work and next steps
├── VERIFICATION.md        # checks and evidence
├── src/                   # reusable functions imported by scripts
├── notebooks/             # exploration and reports
├── logs/                  # run records, reviewed before sharing
└── Snakefile              # optional: dependencies between pipeline steps

Research projects may also need read-only raw storage and a processed-data directory, with access arranged under page 13.

A Snakefile is optional. Start with a script and documented run command; see page 8 when you need more automation.