Home

Welcome to the rai-checklist-cli wiki!

RAI Checklist is a command-line tool to easily add customizable responsible AI checklists to data science, Generative AI, or traditional machine learning projects. This tool helps ensure that AI projects adhere to ethical guidelines and best practices throughout their lifecycle.

It complements the Byte and Atom AI Risk Assessment service - Visit Byte an Atom Research for more

---

Features

• Generate customizable AI responsibility checklists
• Support for various output formats: Markdown (.md), YAML (.yaml), JSON (.json).
• Easily integrate into existing projects or CI/CD pipelines.
• Customizable checklist sections
• Validation of ethical and technical aspects in CI/CD pipelines using YAML or JSON checklists.

New Features Added:

• Support for YAML and JSON: You can now generate checklists in YAML and JSON formats, making it easy to integrate into CI/CD pipelines.
• CI/CD Integration Example: Added GitHub Actions template to automate responsible AI checks.

Installation

Install the Responsible AI Checklist CLI using pip:

pip install rai-checklist-cli

Usage

The basic syntax for using the CLI is:

rai-checklist [OPTIONS]

Options:

• -h, --help: Show help message and exit
• -w, --overwrite: Overwrite existing output file
• -o, --output PATH: Specify output file path
• -f, --format TEXT: Specify output format (md, html, ipynb)
• -l, --checklist PATH: Path to custom checklist file

Examples

Generate a markdown checklist:

rai-checklist -o checklist.md -f md

Generate a YAML checklist:

rai-checklist -o checklist.yaml -f yaml

Generate a JSON checklist:

rai-checklist -o checklist.json -f json

or directly from a jupyter notebook (.md)? try:

# pull down the checklist use --upgrade for the latest 

!pip install rai-checklist-cli
# then creates a markdown file with the sections you care about

!rai-checklist -o checklist.md -s project_motivation problem_definition -f md  

# now read in the markdown content in your notebook

with open('checklist.md', 'r') as f:
    checklist_content = f.read()
from IPython.display import Markdown, display

# and finally display the checklist

display(Markdown(checklist_content))