Starting the client

To interact with the Feature Store, you need to authenticate first. The Python client exposes two entry points:

• login() — the recommended way. It uses H2O AI Cloud discovery to locate the Feature Store services and to authenticate, based on your environment (a Notebook Engine, the H2O CLI configuration, a platform token, or a token provider you supply).
• login_custom() — for advanced or off‑cloud deployments where you provide the service endpoints and OIDC parameters explicitly.

Both return a single FeatureStoreClient instance.

login (platform discovery)

login() discovers the Feature Store and authorization gateway endpoints through H2O AI Cloud discovery and authenticates you. The examples below cover the different ways to provide credentials.

From a Notebook Engine

If you are running the client from within a private Notebook Engine, your environment is already configured — no arguments are required.

Python

import h2o_featurestore
client = h2o_featurestore.login()

Using the H2O CLI configuration

On your local machine, the recommended way is to download and configure the H2O CLI. Visit the CLI & API Access section of H2O AI Cloud to get started. Once configured, calling login() with no arguments picks up the CLI configuration automatically.

Python

import h2o_featurestore
client = h2o_featurestore.login()

If your CLI configuration file lives in a non‑standard location, pass its path explicitly:

Python

import h2o_featurestore
client = h2o_featurestore.login(config_path="/path/to/h2o-cli-config.toml")

Using a platform token

Log in to H2O AI Cloud and visit the CLI & API Access section to generate a platform token. Provide it together with your environment URL:

Python

import h2o_featurestore
client = h2o_featurestore.login(
    environment="https://cloud.h2o.ai",
    platform_token="my-platform-token",
)

Using a token provider

If you already manage authentication yourself (for example, an application that refreshes tokens on your behalf), pass an h2o_authn.TokenProvider. Discovery is still used to locate the service endpoints, but the token provider supplies the access tokens.

Python

import h2o_authn
import h2o_featurestore

token_provider = h2o_authn.TokenProvider(
    issuer_url="https://auth.your-domain.com/auth/realms/your-realm",
    refresh_token="my-secret-refresh-token",
    client_id="oidc-app-client-id",
    client_secret="oidc-app-secret",
)

client = h2o_featurestore.login(token_provider=token_provider)

login parameters

Parameter	Description
environment	H2O AI Cloud environment URL used for discovery. Defaults to the discovered/CLI‑configured environment.
token_provider	An h2o_authn.TokenProvider to authenticate with. When set, it takes precedence over platform_token.
platform_token	Platform token generated from the CLI & API Access page.
config_path	Path to the H2O CLI configuration file when it is not in the default location.
verify_ssl	Whether to verify SSL certificates. Defaults to True.
ssl_ca_cert	Path to a CA certificate bundle for verifying the server certificate.
storage_use_ca_cert	Whether to apply verify_ssl / ssl_ca_cert when connecting to the underlying object storage. Defaults to False.

note

The client configuration is stored by default in the user's home directory. You can change this location by setting the FEATURESTORE_USER_CONFIG environment variable to the desired location before starting the client.

login_custom (explicit OIDC)

Use login_custom() for advanced or off‑cloud deployments where discovery is not available and you supply the Feature Store endpoint, the authorization gateway endpoint, and the OIDC parameters explicitly.

Python

import h2o_featurestore

client = h2o_featurestore.login_custom(
    endpoint="https://featurestore.your-domain.com",
    authz_endpoint="https://authz-gateway.your-domain.com",
    refresh_token="my-secret-refresh-token",
    issuer_url="https://auth.your-domain.com/auth/realms/your-realm",
    client_id="oidc-app-client-id",
    client_secret="oidc-app-secret",
)

login_custom parameters

Parameter	Description
endpoint	Feature Store service URL.
authz_endpoint	Authorization gateway URL used for workspace and user operations.
refresh_token	OIDC refresh token used to obtain access tokens.
issuer_url	OIDC issuer (identity provider) URL.
client_id	OIDC client ID.
client_secret	OIDC client secret. Optional for public clients.
verify_ssl	Whether to verify SSL certificates. Defaults to True.
ssl_ca_cert	Path to a CA certificate bundle for verifying the server certificate.
storage_use_ca_cert	Whether to apply verify_ssl / ssl_ca_cert when connecting to the underlying object storage. Defaults to False.

Using self‑signed certificates

Both login() and login_custom() accept SSL options. To trust a custom CA certificate:

Python

import h2o_featurestore
client = h2o_featurestore.login(
    config_path="/path/to/h2o-cli-config.toml",
    verify_ssl=True,
    ssl_ca_cert="certificate-path",
)

To disable certificate verification entirely (not recommended for production):

Python

import h2o_featurestore
client = h2o_featurestore.login(verify_ssl=False)

Obtaining version

Call client.get_version() to fetch the versions. It returns a ComponentVersions object exposing the client_version and server_version properties (and helpers such as client_is_newer_than_server()).

Python

versions = client.get_version()
print(versions.client_version)
print(versions.server_version)

FeatureStoreClient attributes

Both login() and login_custom() return a FeatureStoreClient instance. The table below lists its primary attributes:

Attribute	Type	Description
workspaces	Workspaces	Create, list, get, and manage workspaces. This is the primary entry point for all workspace and feature set operations. See Workspaces API.
jobs	Jobs	List and get jobs across all accessible workspaces. See Jobs API.
users	Users	Look up users by email. Required before granting feature set permissions. See Permissions.
acl	Acl	Manage feature set permission requests and grants. See Permissions.
projects	Projects	Admin-only migration tool. Retained from v2 to help administrators search legacy projects and migrate them to the workspace model. Not for general use — use workspaces instead. See Admin example.
job_concurrency_limits	JobConcurrencyLimits	Admin-only. Set global and per-workspace job concurrency caps. See Admin example.
dashboard	Dashboard	Access Feature Store dashboard metrics. See Dashboard API.
classifiers	Classifiers	Manage recommendation classifiers. Creating and updating classifiers requires admin access. See Recommendation API.
feature_set_reviews	ClientReviews	List and manage feature set reviews across workspaces. See Feature Set Review API.

---

Feedback

• Submit and view feedback for this page
• Send feedback about H2O Feature Store to cloud-feedback@h2o.ai