Brave Search MCP

Give any MCP client web and local search powered by the Brave Search API.

Freemium· Free: $ 0 USD · Team: $ 4 USD per user/month · Enterprise: $ 21 USD per user/monthMCP Servers

Best for

Developers wiring a lightweight, privacy-respecting web and local search tool into Claude Desktop, Cursor, or a custom MCP agent, and anyone wanting a readable reference MCP server to fork.

Skip if

Teams needing image, video, or news search, streamable-HTTP transport, or an actively maintained connector; they should use Brave's newer official brave-search-mcp-server instead.

Brave Search MCP is a reference Model Context Protocol server, originally maintained by the MCP project, that exposes the Brave Search API to any MCP-compatible client such as Claude Desktop, Cursor, Windsurf, Cline, or a custom agent runtime. It ships two tools: brave_web_search for general web queries with pagination and freshness controls (up to 20 results per call, offset up to 9), and brave_local_search for finding nearby businesses and services with descriptions, ratings, hours, and contact details, plus an automatic fallback to web results when no local matches are found. The server is a small TypeScript/Node package published to npm as @modelcontextprotocol/server-brave-search and distributed as a Docker image, with one-click install buttons for VS Code (NPX and Docker variants) in the README. Typical workflows include grounding an LLM answer in fresh web sources, letting an agent look up a company or restaurant near a user, running quick fact-checks inside an IDE, or wiring Brave results into a RAG or research pipeline as a cheap, privacy-respecting alternative to Google or Bing. Configuration is a single BRAVE_API_KEY environment variable, which you obtain by signing up at api.search.brave.com; the free tier is generous enough for individual use and light agent traffic. Important caveat: the modelcontextprotocol/servers-archived repository was moved to read-only on May 29, 2025, so this specific implementation is no longer maintained by the MCP org. Brave now publishes its own actively developed server (@brave/brave-search-mcp-server / brave-search-mcp on GitHub) with additional tools (image, video, news, summarizer, local POI/descriptions) and both stdio and streamable-HTTP transports, and most users should treat this archived version as a canonical starting point or a stable reference rather than a production dependency.

Editor's take

A clean, minimal reference server that did a lot to popularize MCP in late 2024, but in 2026 it's a museum piece: the archived notice is real and Brave's own maintained server is a straightforward drop-in upgrade with more tools and HTTP transport. Great to read, fine for personal use, don't build a product on it.

— The AI Tool Bible editorial team

Pros

✅ Two focused tools (web + local search) that are easy to reason about and stable to prompt against
✅ Small, single-file TypeScript implementation that's trivial to read, fork, or self-host
✅ Ships as both an npm package and a Docker image, with VS Code one-click install
✅ Backed by Brave's independent index, so results don't depend on Google/Bing rate limits or ToS
✅ Free tier covers 2,000 queries/month, enough for personal agents and prototyping
✅ MIT-licensed reference implementation from the MCP project itself
✅ Local-search fallback to web means agents rarely get an empty result set

Cons

⚠️ Repository is archived (read-only since May 29, 2025) and no longer receives fixes or new features
⚠️ Superseded by Brave's own official server, which adds image/video/news/summarizer tools and HTTP transport
⚠️ Free tier is capped at 1 query per second, which throttles multi-step agents
⚠️ Local search is US-centric and quality varies sharply outside North America
⚠️ No streaming HTTP transport in this version, stdio only, which complicates remote/multi-tenant deployments
⚠️ Brave's index, while good, is smaller than Google's and can miss long-tail queries

Use cases

Grounded web search inside Claude DesktopAgent fact-checking and citation lookupIDE-side research from Cursor or WindsurfLocal business and restaurant discoveryCheap RAG augmentation for small chatbotsMCP server learning and forkingPrivacy-respecting alternative to Google search in agents