Search Japanese hotels by 48 verified kid-friendly conditions (baby beds, kids meals, family baths, etc.), with sources and verification dates. MCP server for kids-stay.com.
Three tools with clear names (list_conditions, search_hotels, get_hotel) and basic descriptions. All tools are READ_ONLY and well-scoped. However, input schemas are not visible in the provided source code, tools.json is referenced but not shown. Parameter descriptions, type constraints, and output schemas cannot be verified. Descriptions are present but brief (10-50 chars each), lacking LLM-optimized detail on WHEN to use each tool and what to expect. No error handling guidance visible. Tool composition is sound (discovery → search → detail), but schema documentation is the critical gap.
Get detailed information about a specific hotel
List all 48 kid-friendly hotel conditions with IDs and descriptions
Search hotels by condition IDs and area/region
Input schemas not visible in source code. tools.json is referenced but not provided. Cannot verify parameter types, constraints, enums, or descriptions.
Tool descriptions are under 50 characters and lack LLM-optimized guidance. 'List all 48 kid-friendly hotel conditions...' does not explain WHEN to call this vs search_hotels, or what structure is returned. Baseline: 194 chars average for A+ tools.
Output schemas not documented in visible code. LLMs cannot plan downstream calls or extract required fields (e.g., does search_hotels return hotel_id for get_hotel?).
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 39 | 2026-07-28+ | v2 |
No error handling guidance visible. If search_hotels returns no results or get_hotel fails, the LLM has no recovery path.
search_hotels and get_hotel descriptions do not clarify parameter requirements or dependencies. E.g., does search_hotels require condition_ids, area, or both? Are they mutually exclusive?