MCP-Web

A Model Context Protocol (MCP) server that provides web search and content extraction capabilities.

Quick Start

• Google API
• Google Search Engine ID
• Jina AI

{
  "mcpServers": {
    "mcp-web": {
      "command": "npx",
      "args": [
        "mcp-web"
      ],
      "type": "stdio",
      "env": {
        "GOOGLE_API_KEY": "XXXXXXXXX",
        "GOOGLE_SEARCH_ENGINE_ID": "XXXXXXXXX",
        "JINA_API_KEY": "jina_XXXXXXXXX"
      }
    }
  }
}

Features

1) DuckDuckGo Search (duckduckgo-search)

Search the web using DuckDuckGo with HTML scraping.

Parameters:

• query (string, required): The search query
• page (integer, optional): Page number (default: 1, min: 1)
• numResults (integer, optional): Number of results to return (default: 10, min: 1, max: 20)

2) Google Search (google-search)

Search Google and return relevant results using the Programmable Search Engine.

Parameters:

• query (string, required): Search query; quotes enable exact matches
• num_results (integer, optional): Total results to return (default: 5, max: 10)
• site (string, optional): Restrict to a specific site/domain (e.g., wikipedia.org)
• language (string, optional): ISO 639-1 language code (e.g., en, es)
• dateRestrict (string, optional): Date filter, e.g., d7, w4, m6, y1
• exactTerms (string, optional): Exact phrase that must appear
• resultType (string, optional): Result type: image|images|news|video|videos
• page (integer, optional): Page number for pagination (default: 1, min: 1)
• resultsPerPage (integer, optional): Results per page (default: 5, max: 10)
• sort (string, optional): Sort order, relevance (default) or date

Note: Requires GOOGLE_API_KEY and GOOGLE_SEARCH_ENGINE_ID to be set.

3) Felo AI Search (felo-search)

AI-powered search with contextual responses for up-to-date technical information (releases, advisories, migrations, benchmarks, community insights).

Parameters:

• query (string, required): The search query or prompt
• stream (boolean, optional): Whether to stream the response (default: false)

4) URL Content Fetcher (fetch-url)

Fetch the clean content of a URL and return it as text with extraction controls.

Parameters:

• url (string, required): The URL to fetch
• maxLength (integer, optional): Maximum content length (default: 10000, min: 1000, max: 50000)
• extractMainContent (boolean, optional): Attempt to extract main content (default: true)
• includeLinks (boolean, optional): Include link text (default: true)
• includeImages (boolean, optional): Include image alt text (default: true)
• excludeTags (array of string, optional): Tags to exclude (default includes script, style, noscript, iframe, svg, nav, footer, header, aside)

5) URL Metadata Extractor (url-metadata)

Extract metadata (title, description, image, favicon) from a URL.

Parameters:

• url (string, required): The URL to extract metadata from

6) URL Fetch to Markdown (url-fetch)

Fetch web pages and convert them to Markdown. Handles HTML, plaintext, and JSON (pretty-printed in a fenced block).

Parameters:

• url (string, required): The URL to fetch and convert to Markdown

7) Jina Reader (jina-reader)

Retrieve LLM-friendly content from a URL using Jina r.reader with optional summaries and formats.

Parameters:

• url (string, required): The URL to fetch and parse
• maxLength (integer, optional): Maximum output length (default: 10000, min: 1000, max: 50000)
• withLinksummary (boolean, optional): Include links summary (default: false)
• withImagesSummary (boolean, optional): Include images summary (default: false)
• withGeneratedAlt (boolean, optional): Generate alt text for images (default: false)
• returnFormat (string, optional): markdown (default) | html | text | screenshot | pageshot
• noCache (boolean, optional): Bypass cache (default: false)
• timeout (integer, optional): Max seconds to wait (default: 10, min: 5, max: 30)

Note: Requires JINA_API_KEY to be set.

Acknowledgments

• Model Context Protocol specification by Anthropic
• DuckDuckGo for providing a privacy-focused web search experience
• Google Programmable Search Engine and Custom Search JSON API
• Jina AI r.reader API for high-quality content extraction
• Felo AI for up-to-date, developer-focused search insights
• Rust ecosystem and crates that power this server:
  • tokio, reqwest, serde, serde_json, tracing, tracing-subscriber, clap
  • scraper, html5ever, markup5ever_rcdom, regex, once_cell, futures, async-stream
  • url, uuid, thiserror, tokio-util, rand, urlencoding
• The broader MCP community for guidance, examples, and discussions

Support

If you encounter any issues or have questions, please open an issue on GitHub.