Brave Search MCP Server
Official MCP server by Brave for web search, local places, videos, images, news, and AI summarization via Brave's independent index.
- Skill Road
- Brave Search MCP Server
Categories
Description
The Brave Search MCP Server is the official Model Context Protocol integration developed by Brave Software. It connects AI agents such as Claude Code, Codex, or Cursor directly to the Brave Search API and its fully independent web index. Rather than relying on search results from legacy search engines or outdated model training checkpoints, language models can use standardized MCP tools to discover current websites, local businesses, news articles, videos, and images, as well as generate automated summaries. The official source repository is maintained by Brave at github.com/brave/brave-search-mcp-server and distributed under the open-source MIT license.
Independent Web Index and Functional Scope
According to the provider, Brave Search operates its own search index built from the ground up, indexing billions of web pages without building profiling data for behavioral advertising. The MCP server exposes this index through multiple dedicated functions.
Universal web discovery is handled by brave_web_search. It conducts general web searches with rich filtering options including pagination, country codes, language filters, and freshness intervals such as the past hour, past day, past week, or custom date ranges. It also supports Brave Goggles to rerank domains based on customized community rules.
Specialized search needs are addressed through dedicated tools. With brave_local_search, agents can locate businesses, places of interest, and addresses with ratings, operating hours, and location data. Timely journalistic coverage is retrieved via brave_news_search, which returns articles with publication timestamps. The brave_video_search function provides video metadata and runtimes, while brave_image_search supplies image URLs and dimensions. In version 2.0, image search shifted to returning clean URL references instead of heavy base64 strings, significantly reducing context token consumption. In addition, brave_summarize allows agents to leverage Brave's built-in AI summarizer to extract key takeaways from search results before adding them into the prompt.
Transports for Local and Hosted Environments
The server architecture supports both local process communication via standard input and output streams and remote network connections via HTTP. In version 2, Brave made STDIO the default operating mode to align with standard MCP conventions in local developer environments. For containerized setups or centralized team infrastructure, switching to HTTP transport is handled by setting BRAVE_MCP_TRANSPORT=http or passing --transport http at startup. Distribution is managed through the npm package @brave/brave-search-mcp-server.
Security Controls and Credential Management
Operation requires a personal API token generated in the Brave Search API dashboard. This credential is provided via the BRAVE_API_KEY environment variable and must never be exposed inside chat prompts or checked into version control. Each search operation sends only the specific search terms and query filters to Brave's endpoints. Local project context, codebase files, and surrounding prompt context remain entirely on the user's system and are not transmitted to Brave.
Licensing and Catalog Assessment
The repository is published under the MIT license. On September 9, 2026, the GitHub API reported exactly 1,432 stars. This figure reflects steady adoption among developers seeking independent search primitives for autonomous agents, though it does not represent an endorsement of security or uptime. Brave offers developer tiers ranging from free introductory request allowances to scalable paid plans for enterprise workloads and local search features. In accordance with Skill Road editorial policies, public catalog entries do not quote fixed prices.
Requirements
Node.js 18 or newer, a personal API key from the Brave Search API dashboard, and the BRAVE_API_KEY environment variable.
Installation instructions
Configure the server via npx: npx -y @brave/brave-search-mcp-server and supply your API key via the BRAVE_API_KEY environment setting.
npx -y @brave/brave-search-mcp-server
Authentication
Authentication is handled via the Brave Search API key generated in the developer dashboard at api-dashboard.search.brave.com.
Required access permissions
Read-only web search access via the Brave Search API; no access to local files outside the process execution runtime.
Transmitted or stored data
Search queries and optional parameters are sent to Brave API endpoints for processing. No ad profiling according to provider statements.
Security risks
Keep the API key confidential in local environment variables. External web results may contain untrusted information and should be evaluated before acting.
License and costs
- License
- MIT
- Cost
- free
The MCP server is MIT-licensed and open source. According to the provider, the Brave Search API offers free starter quotas and scalable usage-based tiers.
Alternatives
Not recorded yet.
At a glance
- Provider
- Brave Software, Inc.
- Status
- Official server
- Deployment
- Local and remote
- Current version
- 2.0.0
- GitHub stars
- 1,432
- Last reviewed
- 09.09.2026
Repository and documentation
Categories
Supported clients
Related guides
Guides and background related to this entry.
Set up Mapbox MCP Server
Set up the Mapbox MCP Server: hosted endpoint or local token, a first test, and sensible limits.
30.09.2026
Set up the Elastic Agent Builder MCP Server
Enable Agent Builder in Kibana, configure tools, and securely connect the built-in MCP endpoint to an AI client via API key or OAuth 2.1.
20.09.2026
Set up the Searchcraft MCP Server
Start Set up the Searchcraft MCP Server with verified links, minimal permissions, and a safe first test.
19.09.2026
Set up Redis MCP Server safely
Configure Redis MCP locally, scope ACL rights, and encrypt the connection.
18.09.2026