MCP.so
ログイン

Kagi MCP Server

@mgomes

基本情報

カテゴリ

検索

ライセンス

MIT

トランスポート

stdio

作者

mgomes

Kagi MCP Server について

Kagi MCP server written in Go

Kagi MCP Server の設定

以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。

{
  "mcpServers": {
    "kagi": {
      "command": "./kagimcp",
      "args": [],
      "env": {
        "KAGI_API_KEY": "YOUR_API_KEY_HERE"
      }
    }
  }
}

Kagi MCP Server のツール

ツール一覧はまだありません

この掲載には表示できるツール一覧がまだありません。サーバーにツールがないという意味ではありません。対応機能はプロジェクトのドキュメントをご確認ください。

ツールは README から自動的に抽出されます。メンテナーは ## Tools という見出しの下に記載することで、このタブに反映できます。

概要

What is Kagi MCP Server?

A Model Context Protocol (MCP) server that integrates with Kagi’s search and summarizer APIs, allowing Large Language Models (LLMs) to search the web and summarize web pages using Kagi’s privacy-focused search and AI capabilities.

How to use Kagi MCP Server?

Install by cloning the repository and building with Go (requires Go 1.18+ and a Kagi API key). Run with -t stdio for subprocess communication or -t sse -port 8080 for an HTTP server. Provide the API key via the KAGI_API_KEY environment variable or the -api-key flag. Integrate with Claude by adding a JSON entry under mcpServers in Claude’s configuration.

Key features of Kagi MCP Server

  • 🔍 Kagi Search: Search the web with Kagi’s privacy-focused search engine
  • 📝 Kagi Summarizer: Summarize web pages using Kagi’s FastGPT APIs
  • 🔄 Multiple Transports: Support for stdio and Server-Sent Events (SSE)
  • 🔑 API Key Management: Flexible options for providing Kagi API keys
  • Two available tools: kagi_search and kagi_summarize

Use cases of Kagi MCP Server

  • Enable an LLM to search the web for current information
  • Summarize any public webpage using AI
  • Integrate web search and summarization into Claude or other MCP-compatible platforms
  • Build privacy-aware research assistants that use Kagi’s search engine

よくある質問

1

What is the Kagi MCP Server and how does it differ from direct API usage?

It provides a standardized MCP interface so LLMs can call Kagi’s search and summarizer APIs as tools, without the LLM needing direct API credentials or custom code.

2

What dependencies or runtime requirements are needed?

Go 1.18+ for building from source, and a Kagi API key (available to Kagi subscribers). No external services beyond Kagi’s APIs are required.

3

Where does data live and what are the transport options?

Data is sent to and from Kagi’s servers. The server supports two transports: stdio (subprocess communication) and SSE (HTTP server, default port 8080).

4

How do I authenticate and configure the API key?

Set the `KAGI_API_KEY` environment variable or pass it via the `-api-key` command-line flag. The key is used to authenticate all requests to Kagi’s APIs.

5

Can I use the Kagi MCP Server with Claude?

Yes. Start the server in stdio mode and add a matching `mcpServers` configuration in Claude’s JSON settings, providing the command and API key.

コメント

「検索」の他のコンテンツ