GetWeb

npm version npm downloads License: MIT GitHub issues

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

Quick Start

• Google API
• Google Search Engine ID
• Jina AI
• Linkup API
• Exa API

{
  "mcpServers": {
    "getweb": {
      "command": "npx",
      "args": [
        "mcp-getweb"
      ],
      "type": "stdio",
      "env": {
        "GOOGLE_API_KEY": "XXXXXXXXX",
        "GOOGLE_SEARCH_ENGINE_ID": "XXXXXXXXX",
        "JINA_API_KEY": "jina_XXXXXXXXX",
        "LINKUP_API_KEY": "XXXXXXXXX",
        "EXA_API_KEY": "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) Linkup Search (linkup_search)

Search the web via Linkup API and return relevant results in Markdown.

Parameters:

• query (string, required): Natural-language search query
• onlySearchTheseDomains (array of strings, optional): Restrict results to specific domains
• dateFilter (object, optional): Date range filter
  • fromDate (string, optional): Start date in YYYY-MM-DD
  • toDate (string, optional): End date in YYYY-MM-DD
• maxResults (integer, optional): Maximum number of results to return (default: 5, min: 1)

Note: Requires LINKUP_API_KEY to be set.

4) Exa Search (exa_search)

Search the web via Exa API and return relevant results in Markdown.

Parameters:

• query (string, required): Natural-language search query
• maxResults (integer, optional): Number of results to return (default: 10, max: 25)
• publishedDateRange (object, optional): Published date range filter
  • fromDate (string, optional): Start date, RFC3339 (e.g., 2024-02-09T00:00:00.000Z) or YYYY-MM-DD
  • toDate (string, optional): End date, RFC3339 (e.g., 2024-02-09T00:00:00.000Z) or YYYY-MM-DD
• crawlDateRange (object, optional): Crawl date range filter
  • fromDate (string, optional): Start date, RFC3339 or YYYY-MM-DD
  • toDate (string, optional): End date, RFC3339 or YYYY-MM-DD
• userLocation (string, optional): Two-letter ISO country code (e.g., US)
• includeText (string, optional): Exact phrase that must appear in the webpage text (max 5 words)
• excludeText (string, optional): Exact phrase that must not appear in the webpage text (max 5 words)
• domain (string, optional): Restrict results to a single domain (e.g., arxiv.org)

Note: Requires EXA_API_KEY to be set.

5) 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)

6) URL Content Fetcher (fetch-url)

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

Parameters:

• url (string, required): The URL to fetch
• maxLength (integer, optional): Maximum content length (default: 30000, min: 1000, max: 500000)
• extractMainContent (boolean, optional): Attempt to extract main content when HTML (default: true)

7) URL Metadata Extractor (url-metadata)

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

Parameters:

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

8) 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

9) 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
• Linkup API for high-quality web search results
• Exa API for fast neural web search
• 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
  • html2text, chardetng, encoding_rs, 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.