unplugin-mcp

Name: unplugin-mcp
Author: situ2001

@situ2001

关于 unplugin-mcp

A unified plugin for developers integrating MCP servers into modern JavaScript build tools, including Webpack, Rollup, Vite, and more.

#webpack-plugin #mcp #rollup #rollup-plugin #unplugin #mcp-server #modelcontextprotocol

基本信息

分类

开发工具

许可证

MIT

运行时

node

传输方式

stdio

发布者

situ2001

配置

使用下面的配置,将此服务器添加到你的 MCP 客户端。

{
  "mcpServers": {
    "unplugin-mcp": {
      "type": "http",
      "url": "http://localhost:14514/mcp/sse"
    }
  }
}

工具

未检测到工具

工具是从 README 中自动提取的。维护者可以在 ## Tools 标题下列出工具,即可填充这部分内容。

概览

What is unplugin-mcp?

unplugin-mcp is a unified MCP (Model Context Protocol) plugin that creates and manages an MCP server, providing AI assistants with tools to analyze your codebase, inspect build configuration, and control the build process. It works with multiple JavaScript build tools through the unplugin abstraction, including Rollup, Vite, Webpack, ESBuild, Rspack, and Rolldown.

How to use unplugin-mcp?

Install the plugin with pnpm add -D unplugin-mcp (or the bundler-specific package like rollup-plugin-mcp). Add it to your build config (e.g., rollup.config.js) and provide built-in or custom tools via the provideUnpluginMcpTools option. The MCP server starts automatically when you run your build tool in watch mode. Then configure your MCP client (e.g., Cursor) to connect to the server’s SSE endpoint, typically http://localhost:14514/mcp/sse.

Key features of unplugin-mcp?

Cross-platform MCP integration for multiple JavaScript build tools.
Bi-directional AI integration for both reading and modifying builds.
Rich built-in tools for module, config, error, and size analysis.
Extensible tool framework via the UnpluginMcpTool interface.
Seamless integration with build tool plugin hooks (e.g., Rollup).
Persistent server that stays alive in watch mode.
Standard HTTP + SSE transport for broad MCP client compatibility.

Use cases of unplugin-mcp?

AI assistant analyzing module dependencies and imports in a Rollup project.
Inspecting current build configuration through a chat interface.
Debugging build errors by asking the AI to examine error details.
Enabling an AI to modify the build process (e.g., adjust plugin options) during development.
Exposing project-specific information to MCP clients via custom tools.