API for annotation queues to programmatically manage them (enqueue, dequeue, status)

[marcklingen]

Describe the feature or potential improvement

Let's track interest for an API for annotation queues. Currently this is an UI-only feature.

Additional information

No response

[dave-mitchell]

I, for one, pictured a workflow like:

1. Trace is scored in the wild (in app/testing)
2. API harvests scores and moves into a queue

[droos1]

I think it woudl be great. And even better being able to set a target ike with llm judge, so when you do a dataset run, they are automatically added to the annotation quueu.