Kagi Translate MCP Server
An MCP server that integrates Kagi Translate as tools for translation, URL translation, and proofreading, requiring a Kagi API key.
README
Kagi Translate MCP Server
An MCP server that exposes Kagi Translate as tools for any MCP-compatible assistant. It is BYOK: set your own KAGI_API_KEY and the server calls Kagi’s Translate API over stdio.
Tools
translate_text- translate plain text or batch text inputstranslate_url- translate content fetched from a URLproofread- proofread and correct text
Requirements
- Node.js 20 or newer
- A Kagi API key in
KAGI_API_KEY
Kagi’s translate service is currently in a free test period. If you need a key, email support@kagi.com.
Setup
npm install
cp .env.example .env
Edit .env and set your real API key:
KAGI_API_KEY=your_real_key_here
Build
npm run build
This compiles TypeScript into dist/.
Run
npm start
For local development with rebuilds:
npm run dev
This command builds once, watches src/ for changes, and restarts the server automatically.
Claude Desktop Configuration
Add the server to claude_desktop_config.json and point it at the built entry point:
{
"mcpServers": {
"kagi-translate": {
"command": "node",
"args": ["/absolute/path/to/kagi-translate-mcp/dist/index.js"],
"env": {
"KAGI_API_KEY": "your_real_key_here"
}
}
}
}
If you prefer to keep secrets out of the config file, start Claude Desktop with KAGI_API_KEY already present in the environment and omit the env block.
Assumptions
The public Kagi docs page is authenticated, so this implementation assumes the confirmed quick-start request shape for translation and conservative optional fields for advanced controls. The code currently sends these payloads:
POST /api/translatewithtextortexts,source_lang,target_lang, and optionalformality,gender, andcontextPOST /api/translate-urlwithurl,source_lang,target_lang, and optionalformality,genderPOST /api/proofreadwithtextand optionallanguageortarget_lang
If Kagi’s authenticated docs differ on field names or response shapes, update src/kagi/types.ts and the tool request mapping in src/tools/.
Implementation Notes
- All outbound Kagi requests go through
src/kagi/client.ts. - Tool inputs are validated with Zod and inferred from the schemas.
- Logging uses
console.erroronly; stdout is reserved for MCP protocol traffic. - Each tool lives in its own file under
src/tools/and is registered fromsrc/tools/index.ts.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。