Single tool with basic schema and minimal description. Tool name 'search_pokemon' follows verb_noun convention (+), but description is only 31 chars, below the 50-200 char baseline for LLM-optimized descriptions. Input schema present with type and description for pokemon_name parameter (+), but output schema is completely undocumented, LLMs cannot predict response structure. No error handling guidance; generic catch-all returns unstructured text. No parameter constraints (e.g., min/max length, format validation). Tool lacks idempotency documentation and does not indicate it is read-only in the description itself.

• Expand tool description to 80-150 chars: 'Search for Pokémon by name and return detailed stats, types, abilities, evolution chain, and habitat. Returns structured data with ID, height, weight, base stats, and image URLs. Call this first to discover available Pokémon before requesting specific details.'
• Document output schema explicitly. Define return structure: { pokemon_id: number, name: string, height_m: number, weight_kg: number, types: string[], abilities: string[], stats: { name: string, value: number }[], evolution_chain: string[], habitat: string, image_url: string }
• Add parameter constraints to pokemon_name: description should state 'Name of the Pokémon (1-50 characters, lowercase or mixed case accepted, e.g. "pikachu", "Charizard"). Partial names not supported, use exact name.'
• Enhance error responses with recovery guidance: 'Pokémon "xyz" not found. Try: (1) Check spelling (e.g., "pikachu" not "pika"), (2) Use lowercase, (3) Call list_pokemon_names() to see available options.' For network errors: 'API temporarily unavailable. Retry in 5 seconds.'
• Add tool annotation in description: 'This tool is read-only and idempotent, safe to retry without side effects.'