TrendWatch exposes 2 tools with basic verb-noun naming (get_trends, get_growth) and reasonable descriptions (60-80 chars). However, critical gaps undermine quality: parameter descriptions are minimal or absent (e.g., 'data_mode' lacks explanation of valid values or purpose), no output schemas are documented, no error handling guidance is provided, and no tool annotations (readOnlyHint, etc.) are present. The tools appear to be thin wrappers around a REST API with minimal LLM-optimization. Descriptions meet minimum length but lack actionable context (e.g., when to use get_trends vs get_growth, what 'data_mode' controls, what fields are returned).

• Document output schemas for both tools. Specify fields returned (e.g., timestamps, values, metadata) so LLMs know what to extract and how to chain calls.
• Expand 'data_mode' parameter description: explain valid values (e.g., 'raw', 'normalized'), when to use each, and default behavior if omitted.
• Add error handling guidance to tool descriptions: 'Raises RateLimited (HTTP 429) if monthly quota exhausted, upgrade at trendsmcp.ai/pricing. Raises TrendsError for invalid keyword or source.'
• Add tool annotations: mark both tools with readOnlyHint=true and idempotentHint=true so LLMs know they are safe to retry without side effects.
• Clarify 'periods' parameter format in get_growth: specify valid period strings ('7D', '30D', '90D', etc.), whether array can be empty, and what the response structure is (e.g., growth per period).