Two tools with minimal parameter schemas (empty objects) and descriptions in Chinese. Tool names follow verb_noun pattern (get_weibo_hot, get_zhihu_hot) but descriptions lack actionable detail for LLM selection. No parameter descriptions, no output schema documentation, no error handling guidance. Input schemas are present but trivial (no properties). Descriptions are 100-120 chars, within acceptable range but generic. No tool annotations (readOnlyHint, etc.). Composition is sound (two independent read-only tools), but definition quality falls short of production baseline.

• Add explicit output schema documentation. Example: 'Returns a formatted string with ranked list of trending topics. Each item includes rank (int), title (string), heat_score (int), and summary (string).'
• Enhance descriptions with WHEN/WHY context. Example for get_weibo_hot: 'Fetch current Weibo trending topics (updated every 5 min). Use this to monitor Chinese social media sentiment, identify viral topics, or track news trends. Returns top 50 ranked by engagement.'
• Implement error categorization in responses. Distinguish: timeout (retryable), network error (retryable), API format change (fatal, contact maintainer), no data (non-fatal, try again later).
• Add readOnlyHint annotation to both tools in Tool definition to signal safety to clients.
• Document result limits and pagination strategy. Current code caps at 50 items, state this in description and explain how to handle pagination if API supports it.

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

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