Three tools with basic Zod schemas and descriptions, but significant gaps in parameter documentation and output schema clarity. All tools have descriptions (10-60 chars, below the 50-200 char baseline for LLM optimization). Parameters lack detailed constraints, format specifications, and dependency documentation. No output schemas documented. Error handling returns generic text responses without recovery guidance. Tool naming is acceptable (verb_noun pattern) but descriptions are too terse to guide LLM selection effectively.

• Expand tool descriptions to 80-150 chars. Example: 'get-weather: Fetch current temperature, conditions, and local time for a city. Use before check-deployment-safety to assess weather impact on deployments. Returns temperature in local units (Celsius for EU/Asia, Fahrenheit for US).'
• Add parameter constraints in descriptions: 'city (string, 1-100 chars, letters/spaces/hyphens only): City name without country code or abbreviations. E.g., "New York", "São Paulo", "Tokyo".'
• Document output schema in tool descriptions or as a separate schema field: 'Returns {content: [{type: "text", text: string}]}. Text field contains formatted weather summary with temperature, conditions, and local time.'
• Enhance error responses with recovery guidance: Instead of 'Error retrieving weather data', return 'Weather API unavailable for {city}, {country}. Verify city name spelling and try again, or check if the city exists in the weather service.'
• Add confirmation step to deploy-to-city: Require a 'confirm' boolean parameter (default false) or implement a two-step pattern: first call returns deployment plan, second call with confirmation ID executes it.