Brave Search
Officialby Brave Software
The Brave Search MCP Server is the official Model Context Protocol integration from Brave Software that connects AI assistants to Brave's independent web search index. Unlike most search APIs that rely on Bing or Google indexes, Brave operates its own large-scale web index built with contributions from its privacy-preserving Web Discovery Project. The server exposes six specialized MCP tools: brave_web_search for comprehensive web searches with advanced filtering and custom re-ranking via Goggles, brave_local_search for discovering nearby businesses with ratings, hours, and contact details, brave_image_search and brave_video_search for rich media discovery with safety filtering and metadata, brave_news_search for current news articles with freshness controls, and brave_summarizer for AI-generated summaries from search results using Brave's summarization API. The server supports both STDIO and HTTP transport modes, making it compatible with all major MCP clients including Claude Desktop, Claude Code, Cursor, Windsurf, and VS Code. Configuration is straightforward with a single BRAVE_API_KEY environment variable, and the free tier provides 2,000 queries per month at no cost. Version 2.x of the server returns response objects that closely mirror the original Brave Search API structure, providing rich metadata including spell check suggestions, content safety classifications, and location-aware results. The package replaces the now-deprecated @modelcontextprotocol/server-brave-search and is actively maintained by Brave Software under the MIT license.
Installation
Key Features
- ✓Six specialized MCP tools covering web search, local business discovery, image search, video search, news search, and AI-powered summarization from Brave's own independent web index
- ✓Privacy-first architecture with Brave's Zero Data Retention policy and an independent search index built partly through the privacy-preserving Web Discovery Project, not reliant on Bing or Google
- ✓Dual transport support with STDIO (default) and HTTP modes, plus granular control via environment variables for port, host, logging level, and tool whitelisting or blacklisting
- ✓Advanced search filtering with custom re-ranking through Brave Goggles, spell check suggestions, content safety controls, and location-aware results with rich metadata
- ✓Free tier offering 2,000 queries per month with zero cost, making it accessible for individual developers and prototyping before scaling to Pro plans
- ✓Multiple installation methods including npx, Docker, Smithery one-click install, and local build from source for maximum deployment flexibility
Use Cases
- →Giving AI coding assistants real-time web search capabilities to look up current documentation, error messages, and solutions without leaving the editor
- →Building privacy-conscious AI applications that require web search without sharing user query data with third-party advertising networks
- →Discovering local businesses, restaurants, and services with ratings, hours, and contact information through natural language queries to AI assistants
- →Creating AI-powered news monitoring and research workflows that retrieve current articles with freshness controls and AI-generated summaries
- →Enriching RAG pipelines with real-time web data from an independent search index for more diverse and comprehensive retrieval results