Three tools with Chinese descriptions and minimal English documentation. Tool names follow verb_noun pattern (list, get, search) but descriptions lack actionable detail for LLM selection. Input schemas present but sparse, only ruleId and keyword parameters with basic type info. No output schemas documented. No error handling guidance. No parameter constraints (enums, ranges). Descriptions average ~80 chars, below the 194-char baseline for production tools. Missing: pagination support, response field documentation, recovery guidance.
按规则 id 获取单条审查规则全文,四段式:适用条件、决策树、禁止写法、正反例
列出团队 AI 代码审查规则库的完整目录:五关各多少条、规则总量、来源构成
按关键词模糊搜索审查规则,返回命中规则的 id、所属关卡和摘要
No output schemas documented. LLMs cannot infer what fields to expect from listReviewRules, getReviewRule, or searchReviewRules responses. Breaks downstream tool chaining and forces LLMs to guess field names.
Descriptions are primarily in Chinese with minimal English context. LLMs trained on English struggle with non-English descriptions. Descriptions lack WHEN/WHY guidance, e.g., searchReviewRules doesn't explain when to use it vs listReviewRules.
searchReviewRules accepts free-form 'keyword' string with no constraints, enums, or format guidance. LLMs may pass invalid keywords or multi-word queries without understanding expected format.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 49 | 2026-07-28+ | v2 |
No pagination support. listReviewRules and searchReviewRules return unbounded results. If rule base grows, responses could exhaust context window. Missing limit, offset, and total_count fields.
No error handling guidance. If getReviewRule receives invalid ruleId, no recovery hint provided. LLM cannot determine whether to retry, search for alternatives, or ask user.