[Vedaant-J]

This PR removes text that tells kumo-engine users to use the hosted service SDK (kumoai_studio / kumoai-studio).

Problem

kumo-engine runs everything in-process and has no dependency on the hosted SDK, but several messages and docs it surfaces pointed users there. For example, a connector naming an identity that does not exist raised:

Connector 'c1' names 'missing' identity, and none exists in this process. Create one and name it on the connector: identity = kumoai_studio.create_identity(CreateIdentityArgs(display_name='<name>', config=SnowflakeIdentityConfig(...), credentials=...)); SnowflakeConnector(identity=identity, ...). ...

A kumo-engine user cannot follow that advice. The same pattern appeared in the duplicate-identity error, connector reader docstrings and doctests (import kumoai_studio as kumo, kumo.LocalBackend), install hints (pip install 'kumoai-studio[s3]'), the kumo.engine connector module docs ("mirrors the cloud SDK's kumoai.connector..."), the demo notebook, and the kumo-api README.

Change

Most of these modules are shared with the hosted SDK, so the text is now library-neutral. Where an example is needed it uses the kumo-engine API (import kumo.engine as kumo, SnowflakeConnector(..., identity=SnowflakeIdentity(...))). The same error now reads:

Connector 'c1' names 'missing' identity, and none exists in this process. Create the identity first (config=SnowflakeIdentityConfig(...), credentials=...), or pass an Identity object on the connector, as in SnowflakeConnector(identity=identity, ...). ...

Files:

• kumo-core: sources/principals.py (two error messages)
• connector readers: kumo-{snowflake,databricks,bigquery,gcs,s3} backend.py, kumo-snowflake spcs.py (docstrings, doctests, install hints)
• kumo-engine: connector/{base,bigquery,databricks,gcs,s3,source_table}.py, config.py, _infer.py, _portable.py (module docs), notebooks/demo_netflix_gnn.ipynb
• kumo-api/README.md

No behaviour change: only message text, docstrings, skipped doctests and markdown change.

Files changed by the open MLOps PRs (#50–#115) are left for a follow-up to avoid conflicts, including kumo_snowflake/artifacts.py, registry.py, sink.py, kumo_databricks/sink.py and kumo/engine/__init__.py.

Testing Done

• kumo-core tests/sources/unit/test_principals.py, test_identity.py: 24 passed
• kumo-snowflake tests/test_spcs.py, test_package_boundary.py, test_identity.py, test_backend.py: 30 passed
• kumo-engine test/test_snowflake_connector.py: 7 passed
• all touched modules import; flake8 clean on touched files

🤖 Generated with Claude Code

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[Vedaant-J]

Change map

File	Function / scope	What changed	Why
packages/kumo-core/src/kumo_core/sources/principals.py	PrincipalStore.create, resolve	error text no longer names kumoai_studio functions	kumo-engine users cannot call them
packages/kumo-snowflake/src/kumo_snowflake/backend.py	SnowflakeSourceBackend docstring	neutral identity wording; doctest uses kumo.engine	stop pointing engine users at the hosted SDK
packages/kumo-databricks/src/kumo_databricks/backend.py	DatabricksSourceBackend docstring	neutral identity wording; doctest uses kumo.engine	same
packages/kumo-bigquery/src/kumo_bigquery/backend.py	BigQuerySourceBackend docstring	neutral identity wording; doctest uses kumo.engine	same
packages/kumo-gcs/src/kumo_gcs/backend.py	GCSSourceBackend docstring	install hint pip install kumo-gcs; doctest uses kumo.engine	same
packages/kumo-s3/src/kumo_s3/backend.py	S3SourceBackend docstring	install hint pip install kumo-s3; doctest uses kumo.engine	same
packages/kumo-snowflake/src/kumo_snowflake/spcs.py	module docstring	example uses SnowflakeConnector.from_spcs_env()	same
packages/kumo-engine/kumo/engine/connector/base.py	module docstring, Connector.sql, table_names, has_table	describe the class directly, no cloud SDK comparison	engine docs should stand alone
packages/kumo-engine/kumo/engine/connector/bigquery.py, packages/kumo-engine/kumo/engine/connector/databricks.py, packages/kumo-engine/kumo/engine/connector/gcs.py, packages/kumo-engine/kumo/engine/connector/s3.py	module and class docstrings	drop kumoai.connector... comparisons; TODOs restated without the SDK	same
packages/kumo-engine/kumo/engine/connector/source_table.py	module docstring, SQLSourceTable	drop kumoai_studio comparisons	same
packages/kumo-engine/kumo/engine/config.py	module docstring	no kumoai_studio.init mention	same
packages/kumo-engine/kumo/engine/_infer.py, packages/kumo-engine/kumo/engine/_portable.py	module and helper docstrings	drop "ported from kumoai/..." notes	same
packages/kumo-engine/notebooks/demo_netflix_gnn.ipynb	markdown cells	describe kumo-engine directly	same
packages/kumo-api/README.md	intro	recommend kumo-engine instead of kumoai-studio	same

[coderabbitai]

Caution

Review failed

The pull request is closed.

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

_{Comment @coderabbitai help to get the list of available commands.}

[Vedaant-J]

Closing: out of scope. kumoai_studio is still the right reference for the hosted SDK, and these shared modules serve both.