Yuque MCP Server - 语雀模型上下文协议服务器,提供对语雀知识库、文档等资源的 MCP 接口访问
This server has basic tool definitions but significant quality gaps across naming, descriptions, and parameter details. 11 of 14 tools lack proper input schema validation (empty or minimal properties). Descriptions are present but often too brief (under 20 chars baseline). Critical parameter descriptions are missing for several tools. No documented output schemas. Error handling is minimal. The server presents as a working Yuque API client wrapper but falls short of production-grade MCP tool quality standards. Language barrier (Chinese docs/descriptions) adds friction for English-speaking agent developers.
创建知识库
删除知识库(谨慎操作)
获取文档详细内容,支持获取完整 Markdown 源码。注意:需要从搜索结果中获取 namespace 和 slug 参数
通过文档ID获取文档内容。注意:语雀API不支持直接通过文档ID获取,此工具会提供使用建议
获取指定版本详情
获取知识库详细信息
根据登录名获取用户信息
get_doc_version has NO documented input schema (empty properties={}) and description under 20 chars. Cannot be reliably invoked by agents.
update_repo and delete_repo use mutual exclusion (repo_id XOR namespace) but do NOT document this constraint in parameter descriptions. Agents cannot know which parameter is required.
All descriptions are in Chinese (e.g. '获取当前语雀用户信息'). English-speaking agent developers cannot understand when to invoke tools without translation. This violates the pattern:tool-description baseline of clear, actionable English prose.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-21 | D | 54 | 2026-07-28+ | v2 |
| 2026-03-09 | D | 59 | 2024-11-05+ | v1 |
获取当前语雀用户信息
查看文档版本历史
列出知识库中的文档
列出指定团队的知识库
列出当前用户的所有知识库
列出指定用户的知识库
更新知识库信息
No documented output schemas for any tool. Agents cannot plan downstream calls or know what fields will be returned. Rubric baseline: 100% of A+ tools have documented return types.
create_repo has 7 parameters but only 'name' is required. No guidance on defaults, mutual exclusion (owner_login vs login), or when each optional param should be provided. Agents will guess.
Destructive operations (delete_repo) lack confirmation/dry-run support. No error guidance. Agents could destroy knowledge bases irreversibly without safeguards.
Parameter descriptions lack format/range constraints. E.g. 'public' in create_repo documents enum [0,1,2] but description does not explain what each means. Agents must infer from enum values alone.
No error handling or recovery guidance. Tools do not document what errors may occur, when to retry, or how to self-correct. Agents will retry blindly on any failure.
get_doc and get_doc_by_id return different data structures for same resource (one uses namespace+slug, other uses doc_id). No guidance on which to use when. Naming also conflates them.