A robust API should have the ability to:

On-demand inspect the backend to see what is active:

• What queues exist, and how many jobs are in them
• What workers are running, and what they're working on
• For a given Job, what configuration was given at enqueue time, when it was enqueued

Inspect history to see what has happened:

• Worker success/failure rates
• Per Job metrics, including at least: success/failure rate, enqueue rate, throughput, and average duration

Listen to an active cluster to get a "real time" view of what's happening:

• Worker started, or quitting
• Job started, and finished