Two tools with minimal descriptions (both under 100 chars) and incomplete parameter documentation. get_task has a vague description that doesn't explain when to use it or what it returns. check_pricing lacks context on what 'pricing' means (cost per call? rate limits?). Both tools have sparse parameter descriptions. No output schemas documented. No error handling guidance. Tool names follow verb_noun pattern but descriptions fail to meet LLM-optimization standards (50-200 chars baseline). No tool annotations (readOnlyHint, etc.) despite both being read-only operations.

• Expand get_task description to ~120 chars: 'Fetch the current status and result payload for a gpt-image task. Returns task_id, status (queued|processing|completed|failed), and output URLs if ready. Call this to poll async task progress.'
• Expand check_pricing description to ~100 chars: 'Look up RunAPI pricing for gpt-image model. Returns cost per call, rate limits, and pricing URL. Use this before creating tasks to estimate costs.'
• Document output schemas in code comments or via a separate schema registry. For get_task, specify: {task_id: string, status: 'queued'|'processing'|'completed'|'failed', task: {id, status, output_urls?, error?}}
• Add parameter descriptions: 'action: Endpoint name where task was created (e.g., generate_image). Required if multiple endpoints exist; optional if only one.' and 'model: Model slug (e.g., gpt-4-vision). Defaults to primary model if omitted.'
• Add error handling guidance in descriptions: 'If task not found, verify task_id is correct. If pricing lookup fails, check model/action names against available options.'