Server defines 6 tools with basic schemas and minimal descriptions. All tools have input schemas with typed parameters (lat/lon as numbers, stationId/q as strings), meeting baseline structure. However, descriptions are critically short (8-16 chars: '장소 검색', '길찾기 계획', etc.), well below the 20-char minimum and far below the 194-char baseline. No output schemas documented. No error handling guidance. Tool names lack English verb prefixes, making them opaque to English-speaking LLMs. No parameter descriptions beyond type names. Composition is reasonable (6 focused tools), but definition quality is severely hampered by missing descriptions and undocumented outputs.

• Expand all tool descriptions to 50 - 200 characters in English. Example: 'search' → 'Search for places (stations, landmarks, addresses) by name and location. Returns place name, ID, and coordinates for chaining to plan().'
• Document output schemas for each tool. Specify returned fields (e.g., search returns {id, name, lat, lon, type}). This enables LLMs to chain tools and extract required IDs.
• Rename tools with action verbs: 'station_arrivals' → 'get_station_arrivals', 'station_timetable' → 'get_station_timetable', 'station_facilities' → 'get_station_facilities', 'barrier_free' → 'search_barrier_free_facilities'.
• Add parameter descriptions: 'lat' → 'Latitude in WGS84 (range -90 to 90)', 'lon' → 'Longitude in WGS84 (range -180 to 180)', 'q' → 'Search query (place name, address, or landmark)', 'stationId' → 'Station ID returned by search() or get_station_*() tools'.
• Add error handling guidance to each tool description. Example: 'If station not found, call search() with the station name. Returns retryable error with retry_after_ms if rate limit exceeded (30 calls/hour per bucket).'