Brave Search MCP Server

Official

Give your AI real-time web search capabilities via Brave Search

searchbraveweb-searchofficialreal-time

The official Brave Search MCP server adds real-time web search to your AI assistant using the Brave Search API. Ask your AI to research topics, find recent news, or look up information that postdates its training cutoff. Results include titles, URLs, and summaries. Requires a free Brave Search API key (generous free tier available). Unlike asking the AI to visit URLs directly, this server searches the web and returns multiple relevant results for the AI to synthesize.

Install Config

Config file: ~/Library/Application Support/Claude/claude_desktop_config.json· Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "brave-search": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-brave-search"
      ],
      "env": {
        "BRAVE_API_KEY": "your_brave_api_key_here"
      }
    }
  }
}

Required environment variables:

BRAVE_API_KEY — Brave Search API key (free tier available)(e.g. BSA...)

Available Tools (2)

Tool name	Description
brave_web_search	Search the web and return ranked results with titles, URLs, and summaries
brave_local_search	Search for local businesses and places

Example Prompts

“Search for the latest news about Model Context Protocol”
“Find recent benchmarks comparing GPT-4 and Claude”
“What are the best restaurants near downtown San Francisco?”
“Search for tutorials on building MCP servers in Python”
“Find the official documentation for Next.js 15”

Pros & Cons

Pros

Real-time web search — not limited to training data
Free API tier available
Returns multiple results for AI to synthesize
Official Anthropic server

Cons

Requires a Brave API key
Search results quality depends on Brave's index
Not suitable for searching private or internal content

Who Is This For?

·Anyone who wants their AI to access current information
·Developers researching libraries, tools, or recent news

When to use: Use when you need current information beyond the AI's training cutoff, or when you want the AI to research multiple sources before answering.