Three tools with complete schemas and descriptions, but quality varies. submit_task has good parameter detail (queue_key, data, context, metadata, summary with maxLength constraint). list_queues and get_queue_config are minimal, descriptions are adequate but lack actionable context for LLM selection. No output schemas documented. Parameter descriptions are present but generic (e.g., 'The queue key/identifier' lacks format/constraint guidance). No error handling guidance visible. Tool names follow verb_noun pattern correctly. Missing: pagination guidance for list_queues, idempotency hints for submit_task, and recovery instructions for failures.

• Document output schemas for all three tools. Example: list_queues returns {queues: [{queue_key: string, review_type: string, assignment_strategy: string}], total: number}.
• Add pagination to list_queues: accept limit (1-100, default 20) and cursor parameters; return next_cursor and total_count.
• Enhance parameter descriptions with format constraints. E.g., queue_key: 'Alphanumeric queue identifier (2-50 chars, lowercase, hyphens allowed).'
• Add error handling guidance to all tools. E.g., submit_task: 'Returns 404 if queue_key not found. Try list_queues() to discover valid queues.'
• Clarify submit_task idempotency: 'Submitting the same task twice with identical data returns the same task_id (idempotent). Modify data to create a new task.'

Inferred effective spec: <=2025-11-25.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry