Brave Search MCP Server

Official

Comprehensive Brave Search capabilities through MCP

How to install ↓ Compare GitHub source repository ↗

Category: Browser #8 of 32
Stars: ★ 1.5k Very popular
Transport: stdio (local process) · Streamable HTTP
Runtime: Node.js 22+ · Docker
Credentials: API key / credential required
License: MIT
Last commit: 22 days ago
Tools: 8

59FMRS · C

This is an official Brave-maintained search MCP Server covering web, news, image, video, local, place, LLM-context, and summarization tools. It requires a Brave Search API key and defaults to STDIO; HTTP deployments require careful handling because the endpoint is unauthenticated.

Strongest · Documentation 15/20 Weakest · Maintenance 7/20

Reliability

9/20

Security and permissions

14/20

Maintenance

7/20

Documentation

15/20

Setup experience

14/20

Why each score

Reliability 9/20: The manifest provides a reproducible npm package, version, stdio transport, and required secret, while the README documents builds and tool parameters, making a normal startup and handshake plausible. However, the supplied material contains no source code, CI, tests, or error-path evidence, so MCP initialization, tool discovery, and implementation consistency cannot be verified. The README also describes HTTP and other modes while the manifest declares only stdio, warranting a conservative deduction.
Security and permissions 14/20: The API key is marked as a required secret and file-based secret loading is documented; HTTP defaults to loopback binding and the README describes unauthenticated exposure, Origin DNS-rebinding protection, and optional Host validation. The server performs search-oriented external calls without delete, payment, or deployment side effects. However, exposed HTTP has no authentication, Host validation is opt-in, and queries and location headers are sent to Brave; stronger permission isolation and data-retention disclosure are not evidenced, so points are deducted.
Maintenance 7/20: The material indicates that the repository is not archived and provides a defined 2.1.0 release and MIT license, establishing basic release and licensing information. It provides no commit history, CI, dependency-update evidence, maintainer response process, vulnerability-reporting path, or security-response channel, so sustained maintenance and governance cannot be established.
Documentation 15/20: The README covers installation, API-key setup, transport configuration, multiple tools and parameter ranges, environment variables, client examples, builds, and Docker usage. It also discloses some Pro-plan limitations and HTTP security considerations. Clear pricing, quotas and rate limits, systematic troubleshooting, test evidence, and implementation-level verification are missing; the manifest and README also differ in how transport support is represented, so points are deducted.
Setup experience 14/20: Npm, Docker, Claude Desktop, and VS Code examples are provided, including secret prompting, default stdio usage, Node.js requirements, and build steps, giving a clear mainstream-client path. Users still need a Brave account and API key, some scenarios require manual building or HTTP configuration, and there is no file-level execution evidence, cross-platform validation, or automated test evidence, so the score is below full marks.

Static review · not runListed 2026-07-29

Read the FMRS scoring method →

Fit and risk

What it can accessUses the network

Best for

AI agents that need Brave Search API data
Workflows requiring web, news, or multimedia search
Applications needing structured place and local-business information

Not for

Offline workflows that do not need an external search service
Environments unable to provide a Brave Search API key
Users needing full local-search capabilities without a Pro plan

Required permissions

Requires a valid BRAVE_API_KEY, or a key file specified through BRAVE_API_KEY_FILE
Requires network access to the Brave Search API
In HTTP mode, host binding and port are controlled by configuration

Risks and side effects

The API key is a required secret and should not be exposed in logs or shared configurations
The HTTP endpoint is unauthenticated; use extra care when binding beyond loopback
Full local-search capabilities and some additional snippet features require a Pro plan
Search results and AI summaries depend on an external service and may require verification

Setup

Before you start

Runtime：Node.js 22+ · Docker

BRAVE_API_KEY requiredsecret Brave Search API key generated from https://api-dashboard.search.brave.com/app/keys; BRAVE_API_KEY_FILE

BRAVE_API_KEY_FILE optionalsecret

Other optional settings (9)

BRAVE_MCP_TRANSPORT optional

BRAVE_MCP_PORT optional

BRAVE_MCP_HOST optional

BRAVE_MCP_ALLOWED_ORIGINS optional

BRAVE_MCP_ALLOWED_HOSTS optional

BRAVE_MCP_LOG_LEVEL optional

BRAVE_MCP_ENABLED_TOOLS optional

BRAVE_MCP_DISABLED_TOOLS optional

BRAVE_MCP_STATELESS optional

Create a Brave Search API account and generate an API key. For npm installation, add the example configuration to Claude Desktop or VS Code and set BRAVE_API_KEY. The server defaults to STDIO; to use HTTP, set BRAVE_MCP_TRANSPORT=http or pass --transport http.

claude_desktop_config.json

{"mcpServers":{"brave-search":{"command":"npx","args":["-y","@brave/brave-search-mcp-server","--transport","stdio"],"env":{"BRAVE_API_KEY":"YOUR_API_KEY_HERE"}}}}

Shown for Claude Desktop. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.

.vscode/mcp.json

{
  "servers": {
    "brave-search": {
      "command": "npx",
      "args": [
        "-y",
        "@brave/brave-search-mcp-server",
        "--transport",
        "stdio"
      ],
      "env": {
        "BRAVE_API_KEY": "YOUR_API_KEY_HERE"
      }
    }
  }
}

Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).

Terminal

claude mcp add brave-search -e BRAVE_API_KEY=YOUR_API_KEY_HERE -- npx -y @brave/brave-search-mcp-server --transport stdio

Run it in a terminal; replace any <…> placeholders with your own values first.

Generate and validate client config

Check that it works

The client's tool list should show the eight brave_* tools including brave_web_search; ask the client to run a web search — returning real results proves the install and API key work.

Troubleshooting

Verify that BRAVE_API_KEY is set, or that BRAVE_API_KEY_FILE points to a key file available inside the container or runtime
Check the npm package version, Node.js 22.x or later, and build completion
Confirm the client configuration uses the correct npm package identifier and STDIO arguments
For HTTP mode, check BRAVE_MCP_TRANSPORT, BRAVE_MCP_HOST, BRAVE_MCP_PORT, and allowed Origin or Host settings
Before using brave_summarizer, run brave_web_search with summary=true and provide the returned summary key

Things to try

Once connected, you can ask your AI assistant things like:

Use brave_web_search to find the latest news about electric vehicles and summarize the results
Find the highest-rated coffee shops near San Francisco with brave_local_search
Use brave_image_search to find high-resolution photos of the northern lights
Generate an AI summary with brave_summarizer using the summary key from my last web search

Tools 8

brave_web_search read-only

Performs web searches with rich result types and advanced filtering options.

brave_local_search read-only

Searches local businesses and places, returning ratings, hours, and AI-generated descriptions.

brave_video_search read-only

Searches videos and returns metadata and thumbnail information.

brave_image_search read-only

Searches images and returns metadata such as URLs, dimensions, and confidence scores.

brave_news_search read-only

Searches news articles with freshness controls and breaking-news indicators.

brave_summarizer read-only

Generates AI-powered summaries from web search results through Brave's summarization API.

brave_place_search read-only

Searches points of interest by geographic area and returns structured data including addresses, hours, contact details, ratings, photos, categories, and time zones.

brave_llm_context read-only

Retrieves pre-extracted web content optimized for AI agents, LLM grounding, and RAG pipelines.

Use cases

Retrieve web information and structured search results

Search news, images, videos, local businesses, and places

Provide web context for AI agents or RAG pipelines

Run a web search with summary=true and then generate an AI summary

Supported clients

Claude Desktop

VS Code

Listed from the project's documentation, not tested by this site.

Overview

Brave Search MCP Server integrates the Brave Search API with MCP, offering web, local business, place, image, video, and news search, plus AI-oriented web context and summarization. STDIO is the default transport, and HTTP is also supported.

Similar servers

Playwright MCP 78 · B

Microsoft's official Playwright-based browser automation MCP server

★ 38.3k · Tools 10 Compare with this →

Chrome DevTools MCP 76 · B

The Chrome team's official MCP server exposing Chrome DevTools capabilities to AI assistants

★ 53.8k · Tools 22 Compare with this →

FSB MCP Server 66 · C

Control the FSB browser extension from any MCP client for DOM-first AI browser automation.

★ 195 · Tools 30 Compare with this →

Playwright MCP Server 66 · C

Playwright-based browser automation MCP server

★ 5.7k · Tools 1 Compare with this →

Compare with other MCP servers

Source revision e4866a959a63 Data synced 2026-10-19 Read the FMRS scoring method