expose comment-assignee and resolution metadata to AI agents in a structured way so I can:

Filter comments where assigned == true and resolved == false.

Group them by comment assignee.