Loading ToolBench
Remote MCP servers for WeRead and Caiyun Weather on Cloudflare Workers.
Server has 21 tools with complete input schemas and descriptions. Naming follows verb_noun pattern (weread_*, caiyun_*, stock_*). Descriptions are present and contextual (avg ~80 chars). However, output schemas are undocumented, tools return JSON via jsonText() helper with no declared structure. Parameters lack validation guidance (e.g., location regex not mentioned in description). Error handling is minimal, upstream failures are wrapped but not categorized for LLM recovery. No tool annotations (readOnlyHint, idempotentHint). Composition is sound: tools are single-responsibility and chainable (search → get_info → get_chapters). Moderate quality, functional but missing production-grade polish.
获取天级天气预报数据。
获取小时级天气预报数据。
获取分钟级天气预报数据。
获取实时天气数据。
获取股票行情数据。
获取全书或某章节热门划线，包含划线原文和划线人数。服务端固定返回前 20 条，不支持分页。
获取单本书划线内容列表。该接口已过滤书签，只返回划线内容和章节信息。
Output schemas undocumented. All tools return JSON via jsonText() helper with no declared response structure. LLMs cannot plan downstream tool calls or extract specific fields without trial-and-error.
No tool annotations. All tools are read-only but lack readOnlyHint annotation. Agents cannot distinguish safe tools from destructive ones without explicit hints.
Error handling lacks recovery guidance. callWeRead() throws generic errors ('WeRead gateway returned HTTP 401') without categorizing as retryable or user-fixable. LLMs cannot self-correct.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 62 | 2026-07-28+ | v2 |
获取书籍章节目录，返回章节 UID、序号、标题、层级、字数、价格和购买状态等。
获取指定书籍的当前阅读进度。progress 是 0-100 的百分比整数，只有 100 表示已读完。
获取书籍基本信息，包括书名、作者、封面、简介、分类、出版社、ISBN、字数和评分信息。
查询指定划线范围下的想法/评论。range 通常来自 weread_book_bestbookmarks 返回值。
获取推荐书籍列表。
获取相似书籍列表。
获取某章节划线热度统计，返回范围、划线人数、热度分数和类型，不包含划线文本。
获取阅读数据统计，包括总阅读时长、阅读天数、平均每天阅读时长、本周阅读时长和本周阅读天数。
查询指定划线范围下的想法/评论。range 通常来自 weread_book_bestbookmarks 返回值。
获取当前用户在单本书上的个人想法与点评，包括划线想法、章节点评和整本书评。
获取单条想法详情，可拉取评论和点赞分页上下文。
搜索微信读书内容。支持全部、电子书、网文小说、有声书/专辑、作者、全文、书单、公众号和文章。
获取当前用户书架。书架总数应按 books.length + albums.length + (mp 非空 ? 1 : 0) 计算。
获取所有有笔记的书的概览。单本书总笔记数按 reviewCount + noteCount + bookmarkCount 计算。
Parameter validation constraints not documented in descriptions. location parameter has regex /^-?\d+(?:\.\d+)?,-?\d+(?:\.\d+)?$/ but description does not mention format. LLMs may pass invalid coordinates.
No pagination result limits documented. weread_search and similar tools accept maxIdx/count but do not state max result count or when to stop paginating. Risk of context window exhaustion.