MCP Server for Cafe AURA reservation and menu management system. Provides tools for AI agents to query cafe information, manage menus with allergen filtering, check seat availability, and manage reservations.
The Cafe AURA MCP server defines 8 tools with mixed quality. All tools have descriptions and input schemas are partially present with Zod types. However, descriptions vary significantly in detail and some lack crucial guidance. Parameter descriptions are inconsistent, some tools have rich, detailed descriptions (e.g., make_reservation with specific constraints and prerequisites), while others are terse (e.g., get_cafe_info and get_wifi_info have zero-parameter tools with no input guidance needed, but missing output schema documentation). Schema validation uses Zod effectively, but output schemas are never explicitly documented, LLMs cannot plan downstream calls without knowing what fields will be returned. Error handling is minimal; tools return JSON but lack actionable recovery guidance. The server targets a specific domain (cafe management) with realistic workflows (search → availability → reserve → confirm), but tool composition lacks sufficient chaining metadata.
予約をキャンセルします。前日18:00までキャンセル無料
Cafe AURAの基本情報(営業時間・住所・電話番号・アクセス・座席数・決済方法)を取得します
Cafe AURAのメニュー一覧を取得します。カテゴリでフィルタ可能(coffee/tea/food/sweets/seasonal)
予約番号で予約内容を確認します
指定日(オプションで時間帯)の席の空き状況を確認します。テラス席は予約不可(先着順)
Cafe AURAのWiFi情報(SSID・パスワード・速度・制限時間)を取得します
Cafe AURAに予約を作成します。水曜定休。テラス席は予約不可(先着順)。前日18:00までキャンセル無料
No output schemas documented. LLMs cannot predict response structure or plan multi-step workflows. Tools return JSON but the response shape is invisible to clients.
Error handling lacks recovery guidance. Tools return JSON error messages (e.g., 'error: reservation not found') but do not suggest next steps or categorize errors as retryable vs. user-fixable.
get_seat_availability description lacks crucial context. Does not explain what 'availability' means (seats remaining? times open?), or what data structure is returned. The optional time parameter is mentioned but not explained for use cases.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 59 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 8 | - | v1 |
条件を指定してメニューを検索します(アレルゲン除外・価格帯・キーワード)
Missing tool-chaining metadata. make_reservation returns a reservation ID, but that ID format is not documented in the response. Similarly, get_reservation accepts a reservationId but no tool returns an example format or guidance on where to find one.
Parameter constraints documented inline but not as formal JSON Schema constraints. date format specified as 'YYYY-MM-DD形式(例: 2026-03-04)' in description text rather than as a regex or format keyword. LLMs parse descriptions inconsistently; formal constraints are safer.