MCP.so
ログイン

Semantic Search MCP Server

@mhart

基本情報

カテゴリ

検索

トランスポート

stdio

作者

mhart

Semantic Search MCP Server について

An MCP Server to search a Vectorize DB

Semantic Search 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": {
    "cloudflare-docs": {
      "command": "npx",
      "args": [
        "mcp-remote@latest",
        "https://mcp.developers.cloudflare.com/mcp"
      ]
    }
  }
}

Semantic Search 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。

上のボタンで一時的なサンドボックスでこのサーバーを実行し、ツール一覧を取得できます。

概要

What is Semantic Search MCP Server?

An MCP (Model Context Protocol) server that provides semantic search over documentation stored in a Cloudflare Vectorize database. It is built using the Agents SDK and powers Cloudflare’s own documentation MCP endpoint.

How to use Semantic Search MCP Server?

The server can be accessed remotely via mcp-remote (e.g., npx mcp-remote@latest https://mcp.developers.cloudflare.com/mcp) in Claude Desktop, or directly in Cloudflare’s AI Playground or VSCode via SSE. For local development, run npm run dev after populating a Vectorize database.

Key features of Semantic Search MCP Server

  • Searches documentation stored in a Cloudflare Vectorize database.
  • Built with Cloudflare’s Agents SDK.
  • Deployable to Cloudflare Workers with one click.
  • Supports remote SSE and stateless MCP connections.
  • Works with Claude Desktop, AI Playground, and VSCode.

Use cases of Semantic Search MCP Server

  • Querying Cloudflare documentation via an MCP-enabled AI assistant.
  • Integrating semantic search into MCP‑compatible chat clients.
  • Running a local development server for testing searches against a custom Vectorize database.

よくある質問

1

What does the server search?

It searches documentation stored in a Cloudflare Vectorize database.

2

How do I use it in Claude Desktop?

Add a `mcpServers` entry with command `npx`, args `["mcp-remote@latest", "https://mcp.developers.cloudflare.com/mcp"]`.

3

Can I test it in Cloudflare’s AI Playground?

Yes. Enter `https://mcp.developers.cloudflare.com/mcp` as the MCP Server URL in the Playground.

4

How do I run it locally?

Run `npm run dev` (requires a populated Vectorize database). Then use `http://localhost:8787/mcp` as the server URL.

5

What transport does it use?

It uses Server‑Sent Events (SSE) for remote access and supports a stateless MCP endpoint at `https://mcp.developers.cloudflare.com/mcp`.

コメント

「検索」の他のコンテンツ