Aviation Weather MCP has 6 well-named tools with consistent verb_noun patterns (get_metar, get_taf, get_pireps, get_route_weather, get_sigmets, get_gairmets). All tools have descriptions (avg ~60 chars, baseline 194). Input schemas are present with typed parameters and enums for constrained inputs (region, hazard, type). However, descriptions are terse and lack context on WHEN to use each tool or what the output contains. No documented output schemas visible. Parameters like 'radius' lack min/max bounds. Error handling and recovery guidance are absent from descriptions.

• Expand tool descriptions to 100-200 chars. Example: 'Get METAR (current weather observation) for an airport. Returns wind, temperature, visibility, and conditions. Use this for real-time weather at a single airport; use get_taf for forecasts or get_route_weather for multi-airport planning.'
• Document output schemas for each tool. Specify fields returned (e.g., wind_speed, temperature, visibility, raw_text) and their types. This enables LLMs to extract and chain data correctly.
• Add min/max constraints to numeric parameters. Example: 'radius (search radius in miles, 1-500, default 50)' instead of just 'radius (search radius in miles, default 50)'.
• Add error recovery guidance to descriptions. Example: 'If airport not found, verify the ICAO code (4 letters, e.g., KJFK). Try search_airports() to find the correct code.'
• Expand enum descriptions. Replace 'conv' with 'convective (thunderstorms)', 'turb' with 'turbulence', 'ice' with 'icing', 'ifr' with 'instrument flight rules (low visibility/ceiling)' so LLMs understand the hazard types.