Five tools with complete JSON Schema definitions and clear descriptions. All tools follow verb_noun naming (weather_get_*). Schemas include proper type constraints (lat/lon ranges, enum units, min/max for numeric parameters). Descriptions are concise (20-60 chars) and action-oriented. However, output schemas are not documented, LLMs cannot predict response structure. Error handling is basic (generic McpError wrapping). No tool annotations (readOnlyHint, idempotentHint). Parameter descriptions are minimal; some lack context (e.g., 'hours' and 'days' lack explanation of what they control). No guidance on when to use each tool vs. alternatives.

• Document output schemas for all tools. Example: weather_get_current returns {temp: number, feels_like: number, humidity: number, description: string, ...}. Include field types and what each represents.
• Add readOnlyHint: true to all tool definitions. This signals to agents that calls are safe to retry without side effects.
• Expand parameter descriptions: 'hours' → 'Number of hours to forecast (1-48; default 48 returns full 2-day forecast)', 'days' → 'Number of days to forecast (1-8; default 8 returns full week)', 'limit' → 'Maximum number of geocoding results to return (1-10; default 5)'.
• Enhance error messages with recovery hints. E.g., 'Invalid coordinates: latitude must be -90 to 90, got 95. Check your input and retry.' or 'API rate limit exceeded. Wait 60 seconds before retrying.'
• Add tool selection guidance in descriptions. E.g., weather_get_current: 'Use for immediate conditions. For forecasts, use weather_get_hourly or weather_get_daily.' This reduces LLM confusion when multiple tools apply.

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

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