The Arato SDK helps developers automate experiment management, data workflows, and notebook creation directly from their IDE or CI pipeline. It creates a smooth bridge between stakeholders by making every experiment reproducible, trackable, and easy to share across the team.
With one SDK run, you can manage data, generate notebooks, execute experiments, and trigger evals from inside your IDE.
Requirements
Python 3.7 or higher
Arato API key
Installation
Install the SDK using pip:
$ pip install arato-client
Environment setup
Create a .env file in the root of your project and add your Arato API key:
ARATO_API_KEY={Arato_API_KEY}How to generate an API key
Go to:
Settings → Token management → Generate new token
Copy the token and add it to your .env file.
Quick Start: Create your first script
Below is a simple script that initializes the client, creates a notebook, and adds a dataset.
from arato_client import AratoClient
import os
# Initialize the client
client = AratoClient(api_key=os.environ.get("ARATO_API_KEY"))
# Create a notebook
notebook = client.notebooks.create(
name="My First Notebook",
description="Getting started with Arato"
)
# Create a dataset
dataset = client.notebooks.datasets.create(
notebook_id=notebook['id'],
name="Test Data",
content=[
{"input": "Hello", "expected": "Hi there!"},
{"input": "How are you?", "expected": "I'm doing well!"}
]
)
print(f"Created notebook: {notebook['id']}")
print(f"Created dataset: {dataset['id']}")
This notebook and dataset will appear automatically in your Arato workspace.
More examples
For more examples check our repository SDK repository: