Loading ToolBench
MCP server for Japanese law information using e-Gov API
Three tools with basic schemas and Japanese descriptions. All tools have input schemas with required parameters and enums where appropriate (search_law category). However, descriptions are minimal (34-52 chars, below the 50-200 char LLM-optimized baseline), parameter descriptions lack depth, and output schemas are completely undocumented. No error recovery guidance. Tools are read-only but lack idempotent/readOnly annotations. The server truncates responses (5000-10000 chars) without documenting pagination or result limits in tool descriptions.
法令番号を指定して法令の詳細データを取得します。
法令の改正履歴を取得します。
法令名や法令番号で法令を検索します。キーワード検索に対応しています。
Descriptions are too short (30-52 chars vs. 50-200 char baseline). 'Get law revision' and 'Get law data' lack context for when/why to use them. LLMs cannot distinguish between similar tools.
Output schemas completely undocumented. Tools return truncated text (5000-10000 chars) but no schema describes fields, structure, or pagination. LLMs cannot plan downstream calls or extract structured data.
Parameter descriptions lack actionable detail. 'keyword' and 'lawNum' have minimal guidance on format, constraints, or examples. 'category' enum values (1-5) are not explained (what does '1' mean? law vs. regulation?).
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 45 | 2026-07-28+ | v2 |
No error recovery guidance. Errors return generic messages ('API request failed: 404') without suggesting next steps. LLMs cannot self-correct or retry intelligently.
Result truncation (5000-10000 chars) not documented in tool descriptions. No pagination parameters (limit, offset, cursor) exposed. Agents cannot retrieve complete results or iterate through large datasets.