Javadoc Mcp
About
Javadoc Mcp is a Model Context Protocol (MCP) server that allows AI assistants to browse JavaDoc documentation using Maven coordinates. It is designed for developers who need programmatic access to Java library documentation via an AI or CLI tool.
Details
- Author
- loong95
- Downloads
- 304
- Categories
- Other
Jump to
- Five MCP tools for browsing JavaDoc: list_packages, list_classes, get_class, get_member, and search.
- Reads from the local Maven repository (~/.m2/repository) or auto-fetches missing -javadoc.jar files.
- Uses cheerio for HTML parsing and adm-zip to read directly from JARs without extracting to disk.
- Implemented in TypeScript with the @modelcontextprotocol/sdk.
Setting up with Highlight
This MCP is not yet compatible with Highlight’s one-click setup. However, you can still use it with Highlight by following these steps:
- Download and install Highlight from highlightai.com/download
- Navigate to the plugins tab and select "Add Custom Plugin"
-
Configure the plugin with the settings below
Plugin Name
Javadoc McpCommand (node, npx, python, etc.)Please refer to the README for specific instructions on how to obtain API keys or other required environment variables.
- Enable "Start Automatically" if you want the plugin to start when Highlight launches
From the repository
Install via npx -y javadoc-mcp or npm install -g javadoc-mcp, then configure your MCP client (e.g., Claude Code or Claude Desktop) with the appropriate command and arguments in settings.json or claude_desktop_config.json. After setup, tools are automatically available to the AI assistant.
Claude Desktop / Cursor
Paste into your MCP client config file to install this server.
{
"mcpServers": {
"javadoc mcp": {
"javadoc": {
"command": "npx",
"args": [
"-y",
"javadoc-mcp"
]
}
}
}
}McpServers
{
"javadoc": {
"command": "npx",
"args": [
"-y",
"javadoc-mcp"
]
}
}javadoc-mcp
一个 MCP (Model Context Protocol) 服务器，让 AI 助手能够通过 Maven 坐标浏览 JavaDoc 文档。功能
提供 5 个 MCP tools，支持逐层浏览 JavaDoc： | Tool | 说明 | |------|------| |list_packages | 列出库中所有包 |
| list_classes | 列出包中所有类、接口、枚举、注解 |
| get_class | 获取类概览文档（签名、描述、成员摘要） |
| get_member | 获取方法/字段的完整文档（参数、返回值、异常等） |
| search | 按关键词搜索类型、成员或包 |
运行要求
- Node.js 18+ - Maven (mvn 需要在 PATH 中可用)
服务会优先读取本地 Maven 仓库中的 JavaDoc JAR；如果缺失，则通过 mvn dependency:get 拉取对应的 -javadoc.jar。
安装
方式一：直接通过 npx 使用
``bash
npx -y javadoc-mcp
`
方式二：全局安装
`bash
npm install -g javadoc-mcp
`
安装完成后，可直接运行：
`bash
javadoc-mcp
`
配置 MCP 客户端
Claude Code
推荐直接通过 npx 启动，这样不需要手动管理安装路径：
编辑 ~/.claude/settings.json：
`json
{
"mcpServers": {
"javadoc": {
"command": "npx",
"args": ["-y", "javadoc-mcp"]
}
}
}
`
如果你已经全局安装，也可以写成：
`json
{
"mcpServers": {
"javadoc": {
"command": "javadoc-mcp",
"args": []
}
}
}
`
Claude Desktop
编辑 claude_desktop_config.json：
`json
{
"mcpServers": {
"javadoc": {
"command": "npx",
"args": ["-y", "javadoc-mcp"]
}
}
}
`
从源码开发
`bash
npm install
npm run build
`
构建产物输出到 dist/ 目录。
调试
使用 MCP Inspector 进行交互式测试：
`bash
npx @modelcontextprotocol/inspector node dist/index.js
`
使用示例
典型的浏览流程：
`
1. list_packages(groupId="org.apache.commons", artifactId="commons-lang3", version="3.14.0")
→ 获取所有包名
2. list_classes(..., packageName="org.apache.commons.lang3")
→ 查看包中的类列表
3. get_class(..., className="org.apache.commons.lang3.StringUtils")
→ 查看 StringUtils 类的概览和方法摘要
4. get_member(..., className="org.apache.commons.lang3.StringUtils", memberName="join")
→ 查看 join 方法的完整文档
5. search(..., query="String", category="type")
→ 搜索名称含 "String" 的类型
`
本地仓库
JavaDoc JAR 直接使用 Maven 本地仓库中的文件，默认路径为 ~/.m2/repository。如果 ~/.m2/settings.xml 中配置了 <localRepository>，则使用该路径。
如需重新拉取某个文档，可删除对应坐标下的 *-javadoc.jar 后再次调用工具。
技术栈
- TypeScript (ES Modules)
- @modelcontextprotocol/sdk — MCP 协议实现
- cheerio — HTML 解析
- adm-zip — 直接从 JAR 读取文件，不解压到磁盘
- zod` — 参数校验Sign in to leave a review
Use Google, GitHub, or an email account so ratings stay tied to real people.
No reviews posted yet.
