MCP server for analyzing project structure, extracting @intent annotations, managing intent packages, and providing semantic capability discovery
Critical failures across naming, descriptions, and schemas. Most tools lack proper input schema definitions visible in source. Descriptions are present but often generic or in Chinese without English translations. Tool names lack consistent verb-noun patterns. Many tools reference archived/retired code paths, suggesting incomplete or abandoned implementation. No evidence of output schema documentation. Error handling is minimal.
向指定 agent 发送一条消息并等待其下一轮回应(send + await 合成)。自动分派通道:该 agent 正在等待回复时,消息作为回答送达;否则作为新消息派发。返回提问(子 agent 需要澄清,继续用 agent_chat 发送你的回答)或最终结果。
生成项目架构视图
检查文件是否符合分层架构规范(行数限制)
清空所有缓存
提取文件及其所有依赖的完整内容(文件级别)
提取文件的 @intent 注释
从选中代码范围提取函数及其直接依赖
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields without knowing response structure.
Descriptions are in Chinese without English translations. LLMs trained primarily on English will struggle to understand tool purpose and selection criteria.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 43 | 2026-07-28+ | v2 |
获取缓存使用统计信息
获取单个意图包的公开视图(含摘要、语义分组、文件映射、跨包引用)。包不存在或已废弃时返回错误。
列出所有可用的 sub-agent。按 skill 分组,返回名称、描述和可用工具。
列举所有可用的意图包名。默认排除已废弃的包。
列出指定架构层的所有 @intent 能力。支持按层级(Data/Application/Adapter)筛选,返回该层级的文件列表及 @intent 语义。
按关键词搜索相关 @intent 能力。支持中英文混合搜索,按匹配度排序。
搜索函数的 @contract 注释
在文件中搜索函数定义,返回完整代码(包含注释)
语义检索意图包。输入自然语言查询,返回匹配的包名、组名和关联理由。
Tool names lack consistent verb-noun patterns. 'analyze_project_structure', 'check_layer_compliance', 'extract_full_context' use different verb styles. Standardize to verb_noun (e.g., get_, list_, search_, create_).
Parameter descriptions are minimal or missing. 'workspaceRoot' and 'directoryPath' appear in multiple tools but lack guidance on format, defaults, or validation rules.
No error handling guidance. Tools throw generic errors without recovery hints. E.g., 'list_layer_capabilities' throws '列出层级能力失败' with no actionable next step for the LLM.