Loading ToolBench
6 tools with complete input schemas and descriptions, but descriptions are domain-specific Japanese text (介護保険, 議事録) that may not optimize for LLM clarity. All tools are READ_ONLY and well-scoped. Naming follows verb_noun pattern (list_, search_, get_). Parameters include enums, optional flags, and limits. However, output schemas are not documented in the visible code, only input schemas are defined. Error handling and recovery guidance are absent. Tool descriptions average ~120 chars, within acceptable range, but lack explicit 'WHEN to use' guidance that would help LLM selection.
議事録の特定議題の原文を取得する。search_minutes の結果から引用するときに使う。
指定市町村が持つ機能（議員・議事録・予算など）を列挙する。
北海道179市町村の一覧を返す。region・機能・active状態で絞り込み可能。横断分析の対象選定に使う。
議員を横断検索する。名前・読み仮名・会派・委員会で部分一致。
北海道全市町村の議事録を横断キーワード検索する（議題単位）。スペース区切りで複数語のAND検索。結果には引用根拠URLが含まれる — 分析結果を出すときは必ず併記すること。
議事録セグメント（発言単位）を検索する。stdio個人利用専用。
Output schemas not documented. Tools return JSON but LLMs cannot infer response structure (fields, types, pagination). Forces LLMs to guess what data is available.
No error handling or recovery guidance. Tools lack descriptions of failure modes (e.g., 'city not found', 'no results'). LLMs cannot self-correct on failures.
search_segments marked 'stdio個人利用専用' (stdio personal use only) but registered as a standard tool. Unclear whether this tool should be conditionally registered or documented as restricted.
Parameter descriptions use Japanese domain terminology (振興局, 議題, 会派) without English equivalents. Non-Japanese-speaking LLMs may struggle to understand constraints.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 60 | 2026-07-28+ | v2 |
get_minutes_excerpt has optional 'agenda_title' parameter but no guidance on when it is required vs. optional. Dependency between 'schedule_index' and 'agenda_title' is undocumented.