Server has 10 tools with explicit schemas and descriptions in Spanish. Naming follows verb_noun pattern (get_, list_, create_, update_, apply_). However, descriptions lack LLM-optimization guidance (many exceed 200 chars), parameter descriptions are minimal or missing context, output schemas are undocumented, and error handling is absent from tool definitions. No tool annotations (readOnlyHint/destructiveHint). Schemas are present but sparse, many parameters lack type constraints, ranges, or validation guidance. This is a typical community server with functional but incomplete definitions.

• Document output schemas for all 10 tools. Example: get_shop_summary should return {shop_name: string, active_products: number, collections_count: number, most_expensive_product: {id, title, price}, most_cheap_product: {id, title, price}}. Include all fields needed for downstream chaining.
• Add parameter constraints: discount_percentage (0-100), limit (1-250), days (1-365), min_price (>0). Use JSON Schema minimum/maximum and pattern fields, and repeat constraints in descriptions.
• Implement error handling guidance in tool descriptions. E.g., 'If collection not found, call list_collections() to see available options.' Add recovery paths for common failures.
• Add tool annotations: mark get_* and list_* tools with readOnlyHint=true; mark create_*, update_*, apply_* with destructiveHint=true. This enables agents to reason about side effects.
• Provide English descriptions alongside Spanish. Translate descriptions to 50-150 chars per tool, focusing on WHAT (action), WHEN (use case), and WHY (benefit). Remove example values; use enums instead.