Loading ToolBench
Both tools have minimal descriptions (Japanese text, ~20-30 chars each) that fail to explain WHEN to use each tool or distinguish between them. Parameter descriptions are present but generic ('optional'). Input schemas are visible with type declarations, but lack enums, format constraints, and validation guidance. No output schema is documented. Error handling returns raw JSON strings without recovery guidance. The tools lack English descriptions, making them inaccessible to most LLM agents. No tool annotations (readOnlyHint, etc.) despite being read-only operations.
すべてのカレンダーからイベントを検索するツール
特定のカレンダータイプからイベントを検索するツール
Descriptions are in Japanese only (~20 chars each) and lack English translations. They do not explain WHEN to use each tool, what distinguishes them, or what the output contains. LLMs cannot select between tools without clear English guidance.
Parameter descriptions are generic ('optional') and do not specify format, constraints, or examples. 'start_date' and 'end_date' lack format guidance (ISO 8601?). 'calendar_type' has no enum of valid values, forcing LLMs to guess or hallucinate.
No output schema documented. Tools return JSON strings with unspecified structure. LLMs cannot plan downstream calls or extract fields without trial-and-error parsing.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 42 | 2026-07-28+ | v2 |
Error responses return raw JSON with generic 'error' field. No recovery guidance (e.g., 'Try search_calendar_events_by_type with a valid calendar_type'). LLMs cannot self-correct.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Both tools are read-only and idempotent, but this is not declared in the schema, forcing LLMs to infer safety.