Genorbex Python SDK

Official Python client for the Genorbex Prompt Registry and OpenAI-compatible model-serving endpoints.

Install

python -m pip install genorbex

Install MLflow support when your environment does not already include MLflow:

python -m pip install "genorbex[mlflow]"

Authenticate

Create an API key at https://www.genorbex.com/settings/api-keys, then configure it as an environment variable:

export GENORBEX_API_KEY="your_api_key"

Managed Genorbex notebooks authenticate automatically and do not require this variable.

For a self-hosted installation, also set:

export GENORBEX_HOST="https://your-genorbex-host.example"

Load and invoke a registered prompt

import mlflow
from genorbex.sdk import ClientSpace

prompt = mlflow.genai.load_prompt(
    "prompts:/main.data_science.sales_summary@production"
)

formatted_prompt = prompt.format(
    number_of_sentences=3,
    sales_data="Revenue increased by 12% in July.",
)

client = ClientSpace().serving_endpoints.get_open_ai_client()
response = client.chat.completions.create(
    model="databricks-claude-sonnet-4-5",
    messages=[{"role": "user", "content": formatted_prompt}],
)

print(response.choices[0].message.content)

The SDK uses only the Python standard library at runtime. Python 3.9 or newer is required.

Direct prompt loading

Applications that do not use MLflow can load the same registered prompt directly:

from genorbex.sdk import ClientSpace

workspace = ClientSpace()
prompt = workspace.prompts.load(
    "prompts:/main.data_science.sales_summary@production"
)

License

MIT