aftr 0.5.0

CLI for bootstrapping Python data projects with UV, mise, and papermill

aftr

    ___    ________________
   /   |  / ____/_  __/ __ \
  / /| | / /_    / / / /_/ /
 / ___ |/ __/   / / / _, _/
/_/  |_/_/     /_/ /_/ |_|

===============================
  A I   f o r   T h e   R e s t
===============================

aftr (AI for The Rest) is a CLI tool for bootstrapping Python data science projects with modern best practices. It combines UV for blazing-fast package management, mise for reproducible tool versions, and papermill for notebook automation.

✨ Features

• 🚀 Fast Setup: Initialize production-ready data projects in seconds
• 📋 Custom Templates: Create and share project templates with custom dependencies and files
• 🔧 Environment Configuration: Interactive setup for AI coding assistants (Claude Code, Codex, Gemini)
• 🔑 SSH Key Management: Generate and configure GitHub SSH keys
• 📊 Data Science Ready: Pre-configured with DuckDB, Polars, Jupyter, and papermill
• 🎯 Opinionated Structure: Clean project layout with data/, notebooks/, src/, and outputs/
• 🔄 Reproducible Environments: mise.toml ensures consistent tool versions across machines

Installation

# Install globally with UV
uv tool install aftr

# Or run directly with uvx
uvx aftr init my-project

Usage

Interactive Mode

Simply run aftr to access the interactive menu:

aftr

Choose from:

• New Project - Create a scaffolded data science project
• Environment Setup - Configure AI CLI tools and SSH keys
• Help - View usage information

Create a New Project

# Interactive prompt for project name
aftr

# Direct project creation
aftr init my-data-project

# Use a specific template
aftr init my-data-project --template garda

# Create in current directory
aftr init my-data-project --path .

Generated project structure:

my-data-project/
├── pyproject.toml          # UV project config with DuckDB, Polars, Jupyter, papermill
├── .mise.toml              # Tool versions (Python, UV)
├── .mcp.json               # MCP server config (Playwright)
├── CLAUDE.md               # AI assistant guidance
├── notebooks/
│   └── example.ipynb       # Sample notebook with papermill parameter tags
├── src/
│   └── my_data_project/    # Python source code (hyphenated names → underscores)
├── data/                   # Input data (gitignored)
└── outputs/                # Results and artifacts (gitignored)

Project Templates

Templates let you customize project scaffolding with different dependencies, files, and configurations. Perfect for team standards or internal packages.

Managing Templates

# List available templates
aftr config list

# Show template details
aftr config show default

# Export default template as a starting point
aftr config export-default -o my-template.toml

# Register a template from a URL
aftr config add https://git.internal.com/templates/garda-py.toml

# Update a template from its source URL
aftr config update garda

# Remove a registered template
aftr config remove garda

Template Format (TOML)

Templates use TOML format with {{project_name}} and {{module_name}} placeholders:

[template]
name = "My Team Template"
description = "Internal data science template"
version = "1.0.0"

[project]
requires-python = ">=3.11"

[project.dependencies]
polars = ">=1.0.0"
internal-utils = ">=0.5.0"  # Internal packages work too

[project.optional-dependencies]
dev = ["pytest>=8.0.0", "ruff>=0.1.0"]

[mise]
uv = "latest"

[directories]
include = ["config", "scripts"]  # Extra directories beyond defaults

[notebook]
include_example = true
imports = ["duckdb", "polars as pl", "internal_utils"]

[files."CLAUDE.md"]
content = '''
# CLAUDE.md
Custom guidance for {{project_name}}...
'''

[files."config/settings.toml"]
content = '''
[database]
host = "internal.db.com"
'''

Template Storage

Templates are stored in platform-specific config directories:

• Linux/macOS: ~/.config/aftr/templates/
• Windows: %LOCALAPPDATA%\aftr\templates\

Environment Setup

Configure your development environment after installation:

# Interactive configuration
aftr setup

# Non-interactive mode (defaults: Claude Code only, skip SSH)
aftr setup --non-interactive

Setup mode features:

• AI CLI Selection: Install coding assistants via bun (Claude Code, Codex, Gemini)
• SSH Key Generation: Create ed25519 keys for GitHub authentication
• Key Display: View and copy existing SSH keys for GitHub setup

Perfect for running after environment bootstrap scripts to complete your dev setup!

Why aftr?

Modern data science projects need more than just a requirements.txt. aftr provides:

1. UV Package Management: 10-100x faster than pip, with built-in virtual environment handling
2. mise Tool Versioning: Pin Python and UV versions in .mise.toml for team consistency
3. Papermill Integration: Notebooks with parameter tags for automated execution
4. AI-Ready Environment: One command to install coding assistants like Claude Code
5. Clean Conventions: Opinionated structure that just works™

Development

cd packages/cli
uv sync
uv run pytest tests/ -v
uv run aftr --help

License

MIT

Links

• GitHub Repository
• Issue Tracker