MCP.so
ログイン

🚀 OneSearch MCP Server: Web Search & Crawl & Scraper & Extract

@yokingma

🚀 OneSearch MCP Server: Web Search & Crawl & Scraper & Extract について

🚀 OneSearch MCP Server: Web Search & Scraper & Extract, Support agent-browser, SearXNG, Tavily, DuckDuckGo, Bing, etc.

基本情報

カテゴリ

検索

ライセンス

MIT

ランタイム

node

トランスポート

stdio

公開者

yokingma

投稿者

Yoking Ma

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": {
    "one-search-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "one-search-mcp"
      ],
      "env": {
        "SEARCH_PROVIDER": "searxng",
        "SEARCH_API_URL": "http://127.0.0.1:8080",
        "SEARCH_API_KEY": "YOUR_API_KEY",
        "FIRECRAWL_API_URL": "http://127.0.0.1:3002",
        "FIRECRAWL_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

ツール

4

Search and retrieve content from web pages. Returns SERP results by default (url, title, description).

Extract structured information from web pages using LLM. Supports both cloud AI and self-hosted LLM extraction.

Scrape a single webpage with advanced options for content extraction. Supports various formats including markdown, HTML, and screenshots. Can execute custom actions like clicking or scrolling before scraping.

Discover URLs from a starting point. Can use both sitemap.xml and HTML link discovery.

概要

What is OneSearch MCP Server?

A Model Context Protocol (MCP) server that integrates with multiple search providers for web search, local browser search, URL discovery, and scraping capabilities using agent-browser. It is designed for developers building AI agents that need to search and extract web content via MCP tools.

How to use OneSearch MCP Server?

Install via npm (npx -y one-search-mcp) or Docker, then configure the search provider and optional API keys via environment variables. Use the exposed tools (one_search, one_scrape, one_map, one_extract) from any MCP client.

Key features of OneSearch MCP Server

  • Supports multiple search engines: SearXNG, Tavily, DuckDuckGo, Bing, Google, Zhipu, Exa, Bocha, and local browser search.
  • Local browser-based search (free, no API keys) using agent-browser with automatic Chromium detection.
  • Tools include: one_search, one_scrape, one_map, and one_extract.
  • Docker image with pre-installed Chromium for zero-setup deployment.
  • one_scrape supports multiple output formats: markdown, HTML, rawHtml, links, screenshot, full-page screenshot.

Use cases of OneSearch MCP Server

  • Perform web searches from an AI agent using a variety of providers with or without API keys.
  • Scrape a single page and return content in markdown, HTML, or screenshot format with optional pre-scrape actions.
  • Discover all URLs from a starting page (link discovery) via one_map.
  • Preprocess content from multiple URLs into clean text blocks for downstream LLM analysis.

FAQ from OneSearch MCP Server

What search providers are supported?

The server supports searxng, duckduckgo, bing, tavily, google, zhipu, exa, bocha, and local (browser-based). Default is local.

Do I need an API key?

It depends on the provider: local and duckduckgo require no API key; searxng requires a self-hosted URL; bing, tavily, google, zhipu, exa, and bocha require an API key.

What browsers are required for local search?

A Chromium-based browser (Google Chrome, Microsoft Edge, Chromium, or Google Chrome Canary) is required. The server auto‑detects installed browsers; otherwise, you can install Chromium via npx agent-browser install.

Can I run OneSearch MCP Server in Docker?

Yes. Docker images are available from GitHub Container Registry and Docker Hub, with Chromium pre‑installed. See the configuration example for Claude Desktop.

How do I migrate from v1.1.0 or earlier?

Remove Firecrawl environment variables, install a Chromium browser, and update to the latest version with npm install -g one-search-mcp@latest. Firecrawl has been replaced by agent-browser for local operations.

コメント

「検索」の他のコンテンツ