A modern Vue.js Model Context Protocol implementation for Vue.js applications. Provides runtime inspection and manipulation of Vue components, router info, and Pinia state management through MCP tools.
Vue MCP Next exhibits major definition quality gaps across all tools. While tool names follow decent verb_noun convention (get_*, edit_*), descriptions are minimal (10-30 chars), parameter schemas lack descriptions entirely, and output schemas are completely undocumented. The server exposes Vue DevTools operations but provides insufficient guidance for LLM tool selection and parameter understanding. Tools like 'edit-component-state' accept complex nested structures (path array, value of any type) with no validation constraints or error guidance. No tool descriptions explain when to use one tool vs another (get-component-state vs get-pinia-state are conceptually similar but undifferentiated). Output structures are never documented, forcing LLMs to guess what fields are returned. This pattern, minimal descriptions, missing schema, no error recovery guidance, is characteristic of developer-facing tools retrofitted as LLM-callable interfaces without human-centered design.
修改指定组件的状态
获取指定组件的状态信息
获取Vue应用的组件树结构
获取Pinia状态管理的数据
获取Pinia stores的树形结构
获取Vue Router的路由信息
在浏览器中高亮显示指定组件
Output schemas completely undocumented. No tool provides a documented return type schema. LLMs cannot predict what fields will be returned, forcing them to guess at downstream tool parameters or require trial-and-error exploration.
Tool descriptions are 10-30 characters, far below the 50-200 character baseline for LLM discernment. Descriptions like '获取Vue应用的组件树结构' (Chinese text, 16 chars) provide minimal context for when/why an LLM should call this tool. No English translation visible; non-English descriptions reduce accessibility and LLM understanding.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 41 | <=2025-11-25 | v2 |
| 2026-03-09 | F | 0 | - | v1 |
Parameter descriptions are missing or insufficient. 'componentName' in get-component-state is labeled '组件名称' (Chinese), no guidance on format, validation, or examples. 'path' in edit-component-state is 'array of strings' with no explanation of what the array represents or how to construct it. LLMs cannot reliably construct valid inputs.
edit-component-state accepts 'value' with type 'any', no constraints on format, range, or allowed types. 'valueType' parameter is optional and poorly documented. No validation guidance for LLMs attempting to modify component state; risk of malformed mutations.
No error handling guidance. Tools do not document what errors can occur, whether they are retryable, or what the LLM should do next. For destructive operations like edit-component-state, no confirmation step or dry-run mode is offered.
Tool disambiguation is absent. get-component-state, get-pinia-state, and get-router-info all retrieve state-like data but their descriptions do not explain which tool to call in which scenario. LLM must guess whether component state is separate from Pinia state.
No pagination or result limiting documented. get-component-tree and get-pinia-tree could return unbounded large structures. No limit parameter, no cursor, no guidance on expected response size.