Three tools with complete schemas and descriptions, but significant gaps in parameter documentation and error guidance. Tool names follow verb_noun convention (log_meal, calculate_carbs, get_meals). Descriptions are present (avg ~80 chars) but lack actionable context for LLM selection. Parameter descriptions exist but are minimal (e.g., 'Description of the meal eaten' lacks format/constraint guidance). No output schema documentation visible. Error handling returns generic error messages without recovery guidance. Missing tool annotations (readOnlyHint, destructiveHint). The server uses AI-based carb calculation via a sampling client, but this dependency is not documented in tool descriptions.

• Add output schema documentation to each tool. Example for log_meal: 'Returns {id: string, description: string, timestamp: ISO8601, foods: [{name, quantity, carbs}], totalCarbs: number, confidence: string, createdAt: ISO8601}'.
• Enhance parameter descriptions with format and constraint details. Example: 'description (required, 1-500 chars): Natural language description of the meal, e.g., "grilled chicken breast with rice and broccoli"'.
• Implement actionable error messages with recovery hints. Example: 'Invalid timestamp format. Expected ISO 8601 (e.g., 2024-01-15T14:30:00Z). Retry with corrected timestamp or omit to use current time.'
• Add tool annotations: log_meal with destructiveHint=true; calculate_carbs and get_meals with readOnlyHint=true. This enables agents to reason about side effects.
• Document the AI-based carb calculation dependency in log_meal description: 'Uses AI to analyze meal description and estimate carbohydrate content. May ask clarifying questions if description is ambiguous.'