Data Science Project Foundation

Reusable, editor-agnostic template for data science projects with:

• uv as the Python package manager
• generated requirements.txt alongside uv.lock
• centralized virtual environments in C:/Users/Public/.venvs
• portable project standards for VS Code, Positron, Cursor, Claude Code, and Copilot

Prerequisites and Availability

• uv must be installed and reachable on PATH.
• The Python-first bootstrap (python scripts/bootstrap.py) requires a local Python interpreter to launch the script.
• uv can still manage project/interpreter environments after startup.
• In restricted environments without a preinstalled Python interpreter, use the fallback bootstrap path documented in PROJECT_SETUP_README.md.

Quick Start

1. Copy this template into a new project folder.
2. Run python scripts/bootstrap.py.
3. Add dependencies with uv add <package>.
4. Regenerate requirements with python scripts/export_requirements.py.
5. Use the tasks in .vscode/tasks.json or scripts in scripts/.

For a full setup and daily-usage checklist, see PROJECT_SETUP_README.md.

Core Rules

• Do not create .venv in project root.
• Keep uv.lock committed.
• Treat requirements.txt as generated output.
• Keep operating guidance in AGENTS.md.

Layout

• AGENTS.md: canonical workflow and context file
• docs/standards/: source-of-truth standards (tool agnostic)
• .cursor/: optional Cursor adapters (rules + skills)
• .vscode/: VS Code and Positron settings/tasks
• .github/: CI and Copilot instructions
• scripts/: bootstrap, export, and validation utilities