The server defines 3 tools with complete input schemas and descriptions. All tools follow verb-noun naming convention (get_*, calculate_*). However, descriptions are generic and lack context for agent selection. Parameter descriptions are minimal or absent. Output schemas are not documented. Error handling is basic, errors return raw messages without recovery guidance. The server is functional but lacks LLM-optimized descriptions and comprehensive parameter documentation expected of production-grade tools.

• Add documented output schemas for each tool. Example for get_products: 'Returns an object with fields: products (array of {id, name, type, caffeine, dual_transporter, prices_by_region}), total_count (int), filters_applied (object).'
• Expand tool descriptions to 100 - 200 characters, including when to call it. Example: 'Returns Lecka products filtered by type, caffeine, or dual-transporter status. Use to discover available products before calculating nutrition plans. All filters are optional, omit to see entire catalog.'
• Add descriptions to every parameter. For get_products: 'type: Product category. Must be one of: gel, ultra_gel, bar, variety_pack. Omit to return all types.' For calculate_fueling_plan: 'weight_kg: Athlete body weight in kilograms (30 - 150). Used to calculate carb and sodium targets.'
• Add tool annotations. In toolDef schemas, set readOnlyHint: true for all three tools to formally declare they do not mutate state.
• Improve error messages to guide recovery. Instead of 'Error: product not found', return 'Product ID not found. Call get_products() to list available products.'
• Add range constraints to numeric parameters: goal_minutes (30 - 1440), weight_kg (30 - 250). State in both description and as min/max in JSON Schema if supported.
• Document if result limits are applied. If get_products returns ≤100 products, state this in description: 'Returns up to 100 products. Use filters (type, caffeine, dual_transporter) to narrow results.'