Skip to content
MCP server

MCP Web Research

Brave-backed web search plus SSRF-checked page fetching, extraction and comparison

Runs locallyStatus and history
Install
Pricing
Open source
Platforms
CLI, API
Runs
Locally
Version
0.1.0
Listing updated
today

Similar MCP servers

About MCP Web Research

MCP Web Research is an open-source MCP server for web search and scraping. It runs locally on your machine and is started by your AI client.

Web Research mixes search with careful page reading. Two tools need a search provider, while four page tools work without one: fetching a URL as clean text or links, pulling the section around named fields such as pricing or contact, comparing two to five pages around a focus area, and a server_info call.

No language model is involved; research_topic and compare_pages arrange extracted text rather than generating it. Every fetch is checked against server-side request forgery rules, including each redirect hop, with a ten second default timeout and a one megabyte cap on page bodies.

Search uses the Brave Search API, which has a free tier; a DuckDuckGo scraper is available only by explicit opt-in. Without a key the search tools return an error naming the variable to set.

Install

Run in a terminal

claude mcp add mcp-web-research --env BRAVE_SEARCH_API_KEY='<BRAVE_SEARCH_API_KEY>' -- npx -y @opafex/mcp-web-research

Add to claude_desktop_config.json

{
  "mcpServers": {
    "mcp-web-research": {
      "command": "npx",
      "args": [
        "-y",
        "@opafex/mcp-web-research"
      ],
      "env": {
        "BRAVE_SEARCH_API_KEY": "<BRAVE_SEARCH_API_KEY>"
      }
    }
  }
}

Add to ~/.cursor/mcp.json

{
  "mcpServers": {
    "mcp-web-research": {
      "command": "npx",
      "args": [
        "-y",
        "@opafex/mcp-web-research"
      ],
      "env": {
        "BRAVE_SEARCH_API_KEY": "<BRAVE_SEARCH_API_KEY>"
      }
    }
  }
}

Add to .vscode/mcp.json

{
  "servers": {
    "mcp-web-research": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "@opafex/mcp-web-research"
      ],
      "env": {
        "BRAVE_SEARCH_API_KEY": "<BRAVE_SEARCH_API_KEY>"
      }
    }
  }
}

Add to ~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "mcp-web-research": {
      "command": "npx",
      "args": [
        "-y",
        "@opafex/mcp-web-research"
      ],
      "env": {
        "BRAVE_SEARCH_API_KEY": "<BRAVE_SEARCH_API_KEY>"
      }
    }
  }
}

Add to your client's MCP settings (most use this shape)

{
  "mcpServers": {
    "mcp-web-research": {
      "command": "npx",
      "args": [
        "-y",
        "@opafex/mcp-web-research"
      ],
      "env": {
        "BRAVE_SEARCH_API_KEY": "<BRAVE_SEARCH_API_KEY>"
      }
    }
  }
}

Using several servers? Build one config file with the MCP config generator.

Configuration

  • BRAVE_SEARCH_API_KEYBrave Search API key; enables web_search and research_topic (recommended)
  • WEB_RESEARCH_SEARCH_PROVIDERSearch provider: "brave" or "duckduckgo" (unofficial HTML scraper, explicit opt-in only)
  • USER_AGENTUser-Agent sent with every request (default identifies the bot and links to the repo)
  • SEARCH_CACHE_TTLSearch result cache TTL in seconds (default: 900)
  • PAGE_CACHE_TTLFetched page cache TTL in seconds (default: 3600)
  • FETCH_TIMEOUT_MSPer-request timeout in milliseconds, across all redirect hops (default: 10000)
  • MAX_HTML_SIZEMaximum bytes read from a page body (default: 1000000)

Tools this server provides (6)

Taken from the project documentation; the running server may differ.

web_search
Search the web
fetch_webpage
Fetch a URL as clean text or links
research_topic
Search a topic and compile key sentences from top results
extract_structured_data
Return page sections around requested fields
compare_pages
Compare 2 to 5 pages
server_info
Report server information

What it does

Web search
Returns title, URL and snippet per result via Brave.
Page fetching
Returns clean text, links, or both for a URL.
Topic research
Searches, fetches top results and compiles key sentences by source.
Field extraction
Returns the section around requested fields like pricing.
Page comparison
Compares two to five pages side by side.
SSRF protection
Checks every request and redirect hop.

Good for

  • Gather sources on a topic without an LLM in the middle
  • Pull pricing sections from several vendor pages
  • Compare two competitor pages on one theme
  • Read a page as clean text for an assistant

Pricing

Open source

Free server; Brave Search API has a free tier.

MCP Web Research FAQ

Does it need an API key?

Only for search; the page tools work without one.

Which search providers are supported?

Brave, and DuckDuckGo by explicit opt-in.

How long are results cached?

Search results 900 seconds and pages 3,600 by default.

Is MCP Web Research free?

Yes. MCP Web Research is open source, so you can use and self-host it for free.

How do I add MCP Web Research to Claude Code?

Run this command in your terminal: claude mcp add mcp-web-research --env BRAVE_SEARCH_API_KEY='<BRAVE_SEARCH_API_KEY>' -- npx -y @opafex/mcp-web-research

What are the best alternatives to MCP Web Research?

The closest alternatives to MCP Web Research are Scrapling MCP Server, Firecrawl MCP Server, Douyin_TikTok_Download_API, SearXNG Search and machine_internet.

Where this comes from

  • Description: written from the project's README.
  • Install steps: from the official MCP Registry entry.

How AIPoweredBot checks servers