About datarobot_early_access

DataRobot is a client library for working with the DataRobot platform API. This package is the “early access” version of the client. Do NOT use this package in production–you will expose yourself to risk of breaking changes and bugs. For the most stable version, see the quarterly release on PyPI at https://pypi.org/project/datarobot/.

This package is released under the terms of the DataRobot Tool and Utility Agreement, which can be found on our Legal page, along with our privacy policy and more.

Installation

Python >= 3.7 are supported. You must have a datarobot account.

$ pip install datarobot_early_access

Usage

The library will look for a config file ~/.config/datarobot/drconfig.yaml by default. This is an example of what that config file should look like.

token: your_token
endpoint: https://app.datarobot.com/api/v2

Alternatively a global client can be set in the code.

import datarobot as dr
dr.Client(token='your_token', endpoint='https://app.datarobot.com/api/v2')

Alternatively environment variables can be used.

export DATAROBOT_API_TOKEN='your_token'
export DATAROBOT_ENDPOINT='https://app.datarobot.com/api/v2'

Extra

datarobot_early_access has the following optional groups:

• application-utils (requires Python 3.11+): Light async ORM over the

  DataRobot Agentic Memory Service, plus an AG-UI chat-history layer (models, repositories, and AG-UI event storage). This can be used in DataRobot Custom Applications and Agent Workflows.

• auth (requires Python 3.9+): Provides an abstraction to handle OAuth2 authentication with DataRobot API (11.1+).

  This can be used in DataRobot Custom Applications and on its own.

• auth-authlib (requires Python 3.9+): OAuth2 authentication handling via Authlib.

  This can be used in DataRobot Custom Applications and on its own.

• core (requires Python 3.8+): Platform library functions to improve building with DataRobot.

  This can be used in DataRobot Custom Applications, Custom Models, and Agent Workflows.

• fs (requires Python 3.9+): Provides file system implementation to work with the DataRobot file system.

• pipelines (requires Python 3.10+): Adds the runtime needed to author pipelines locally

  (the dr.task / dr.pipeline decorators and dr.CloudExecutor). The pipelines REST surface (dr.Pipelines, dr.Pipeline, dr.PipelineImage, …) is in the base install and needs no extra.

• query-engine (requires Python 3.9+): Engine to run SQLAlchemy-compatible queries against DataRobot data stores

  and JDBC connections.

You can install these optional groups by specifying them in the pip command, for example:

$ pip install datarobot_early_access[auth]

Helpful links

• API quickstart guide

• Code examples

• Common use cases

Bug Reporting and Q&A

To report issues or ask questions, send email to the team.