create-ml-app

A CLI tool that scaffolds a complete ML project from scratch. Type a few things, hit enter, and you have a working ML app.

Made by Monster.

Usage

npx create-ml-app my-project

You will be asked a few questions — project name, type, framework, and whether you want a cookbook recipe. Then it generates everything and installs dependencies.

What you get

• src/ with training, prediction, and config code
• data/ and models/ folders for your datasets and saved models
• tests/ with a basic test suite
• cookbook/ with 6 ready-to-use ML recipes
• package.json preconfigured with the right dependencies
• .gitignore, .npmrc, Dockerfile, and more

Cookbook recipes

Recipe	What it does
Image Classifier	CNN for classifying images into categories
Sentiment Analyzer	Text classification with DistilBERT
Sales Predictor	Neural net for forecasting sales numbers
Fraud Detector	Binary classification for fraud detection
Stock Forecaster	LSTM model for time series prediction
Text Summarizer	Abstractive summarization with DistilBART

Each recipe comes with sample data and real working code — not placeholders.

Framework options

• TensorFlow / Keras — @tensorflow/tfjs-node
• PyTorch — Python-based, cookbook recipes provided
• Scikit-learn — Python-based, cookbook recipes provided
• XGBoost / LightGBM — Python-based, cookbook recipes provided
• Hugging Face Transformers — @huggingface/transformers

Advanced settings

Toggle these on during scaffolding:

• Test suite (Jest + Vitest)
• Docker setup
• GitHub Actions CI/CD
• Pre-commit hooks
• README shields badges
• .env.example

Project structure

my-project/
├── src/
│   ├── index.js
│   ├── train.js
│   ├── predict.js
│   └── config.js
├── data/
├── models/
├── tests/
├── config/
├── cookbook/
├── package.json
├── .gitignore
├── .npmrc
└── Dockerfile

Next steps

cd my-project
npm run train
npm run predict

Installation

# Quick start
npx create-ml-app my-project

# Or install globally
npm install -g create-ml-app
create-ml-app my-project

Contributing

Fork the repo, make your changes, and open a pull request.

License

MIT — Monster