The Lume Python SDK is designed to be simple and intuitive. The API is centered around a few core functions and a primary data object, LumeRun.

This reference is divided into the following sections:

  • Functions: Top-level functions like lume.run() for interacting with the Lume platform.
  • Classes: Core data objects, primarily the LumeRun object that represents a pipeline execution.
  • Exceptions: Custom exceptions raised by the SDK for specific error conditions.

​
Authentication

The SDK authenticates automatically by reading the LUME_API_KEY from your environment variables. You can also initialize the client explicitly using the lume.init() function, which is useful in environments where you cannot set environment variables.

​
Basic Workflow

The most common workflow involves two primary steps:

  1. Calling lume.run() to trigger a new pipeline.
  2. Using the returned LumeRun object to monitor the execution, either by blocking with run.wait() or by periodically checking its status.
import lume

# Trigger a run
run = lume.run(
    flow_version="my-flow:v1", 
    source_path="unique-batch-id-123"
)

# Wait for it to complete
run.wait()

# Check the final status
print(f"Run {run.id} finished with status: {run.status}")