model-registry-client 0.1.0

Python client for the Kubeflow Model Registry REST API

Model Registry Client

Python client for the Kubeflow Model Registry REST API.

Installation

pip install -e ./model_registry_client
pip install "git+https://example.com/your-org/mlops-model-registry.git#subdirectory=model_registry_client"

Usage

from model_registry_client import ModelRegistryClient

registry = ModelRegistryClient(
    base_url="http://mlops-edgemodule.model-registry.svc.cluster.local:80/model-registry/api",
)

model = registry.register_model(
    "my-model",
    "/path/to/model-artifacts",
    version="2.0.0",
    description="Experiment checkpoint",
    metadata={"framework": "pytorch"},
    artifact_type="weights",
)

print(model.id)
print(model.model_name)
print(model.versions)

All resource responses are typed objects with attribute access and a to_dict() method for raw JSON-style data.

Documentation

The package includes MkDocs documentation generated from code docstrings and type annotations:

cd model_registry_client
pip install -e ".[docs]"
mkdocs serve

Build the static site with:

mkdocs build

LLM Markdown Reference

Generate a single Markdown API reference for LLM prompts or retrieval:

cd model_registry_client
pip install -e ".[llm-docs]"
pydoc-markdown pydoc-markdown.yml

The generated file is written to llm_docs/api_reference.md.