Two tools with verb-noun naming (get_week_summary, get_day_details) and basic descriptions present. However, parameter descriptions are incomplete or missing type clarity. Input schemas visible but lack full JSON Schema rigor (e.g., 'boolean|number' is not valid JSON Schema type union). Output schemas are not documented, LLMs cannot predict response structure. No error handling guidance. Tool descriptions are adequate (100-150 chars) but parameters lack actionable constraints. No enum declarations for include_sessions despite it accepting specific modes. Overall structure is present but falls short of production quality.

• Document output schemas for both tools. Specify all returned fields, types, and nesting. Example: {date_start: string, date_end: string, week_shape: {shape_ctl: number|null, acwr: number|null}, salient_sessions: [{id, date, type, duration_minutes, training_load}, ...], ...}
• Replace 'boolean|number' with proper JSON Schema: use oneOf or enum. Define include_sessions as enum: ['all', 'none'] or numeric threshold with min/max bounds (e.g., 0 - 1000 for training load).
• Add parameter constraints to descriptions. Example: 'include_sessions: Pass true for all sessions, false for none, or a numeric training load threshold (0 - 1000) to filter by intensity.'
• Add error handling guidance. Document what happens if year/week_number are invalid, if date is out of range, or if now_iso_date is malformed. Provide recovery hints: 'If week_number is invalid, call get_current_week() first.'
• Clarify tool selection in descriptions. Example: 'Use get_week_summary for overview of a full ISO week; use get_day_details for granular metrics on a specific date.'