Three tools with minimal documentation. All have descriptions (10-50 chars, below the 194-char baseline), no visible input schemas in the provided code, and no output schema documentation. Tool names follow verb_noun convention (buscar_, calcular_, obtener_) but descriptions are generic and lack context for LLM selection. No parameter descriptions visible. No error handling guidance. No evidence of idempotency, pagination, or composition patterns.

• Add full JSON Schema to each tool with type, enum, minLength, maxLength, pattern, and description for every parameter. Example: 'query' should specify max length (e.g., 100 chars) and whether it supports wildcards.
• Expand descriptions to 100-200 chars. For buscar_productos: 'Search the product catalog by name or category. Returns up to 20 matching products with ID, name, price, and category. Use this to discover available items before calculating totals.'
• Document output schemas. Specify what fields each tool returns (e.g., buscar_productos returns [{id, name, price, category, stock}]; calcular_precio returns {total_price, vat, items_count}).
• Add parameter descriptions with constraints. E.g., 'producto_ids: Array of product IDs (1-100 items). Each ID must be a positive integer. Example: [1, 5, 12].'
• Implement error handling with recovery hints. E.g., 'No products found for query "xyz". Try a shorter query or browse categories: electronics, furniture, office-supplies.'