MCP Server

OfficialPrivacy-FirstFree Tier

Brave Search MCP Server

Q: What is the Brave Search MCP Server?

Brave Search MCP Server is a Model Context Protocol integration that gives AI coding assistants access to real-time web search results through the Brave Search API. It lets Claude Code, Cursor, and other MCP clients search the web, get current information, and look up local business data — all without leaving your coding session.

Q: How much does the Brave Search API cost?

Brave Search API offers a free tier with 2,000 queries per month — more than enough for individual developer use. Paid plans start at $5/month for 20,000 queries. For most MCP use cases, the free tier covers daily development workflows. No credit card is required to start.

Q: How do I set up Brave Search MCP with Claude Code?

Run: claude mcp add brave-search -- npx @anthropic/mcp-server-brave-search. Set your BRAVE_API_KEY environment variable with a key from brave.com/search/api. Claude Code will then have access to web search and local search tools.

Q: What is the difference between web search and local search?

Web search returns general web results — articles, documentation, Stack Overflow answers, blog posts. Local search returns business listings with addresses, phone numbers, ratings, and hours of operation. The MCP server exposes both as separate tools, and your AI assistant chooses the right one based on your query.

Q: Does Brave Search MCP track my queries?

Brave Search does not profile users or sell data. API queries are subject to Brave Search API terms, which are separate from the browser privacy policy. However, because queries go through the Brave API servers, the search terms are transmitted to Brave. For maximum privacy, review the Brave Search API privacy policy.

Q: Can I use Brave Search MCP for research tasks?

Yes. Brave Search MCP is ideal for research workflows: competitive analysis, technology comparisons, finding documentation, checking current pricing, and gathering market data. The AI assistant can run multiple searches, synthesize results, and produce structured reports — all in a single conversation.

by Anthropic · brave.com/search/api

Brave Search MCP Server brings real-time web search into your AI coding assistant. It connects Claude Code, Cursor, and Codex to the Brave Search API — giving your agent access to current web results, documentation, pricing data, and local business information without opening a browser.

Unlike AI assistants with static training data, an agent with Brave Search MCP can check current documentation versions, verify API pricing, research competitors, and find the latest solutions to technical problems. It turns your assistant from a knowledge snapshot into a live research partner.

2K/mo

Free Queries

no credit card

Result Types

web + local

<2min

Setup Time

npx one-liner

No Tracking

Privacy

Brave policy

Quick Install

claude mcp add brave-search -- npx @anthropic/mcp-server-brave-search

Key Features

Real-Time Web Search

Search the web for current information — documentation, pricing, changelogs, Stack Overflow answers, and technical blog posts. Results include titles, URLs, and snippets.

Local Business Search

Look up businesses by location — addresses, phone numbers, ratings, hours, and reviews. Useful for building location-based features or market research.

Privacy-First Architecture

Brave Search does not profile users, build search histories, or sell data. Your research queries stay between you and the Brave API.

Generous Free Tier

2,000 free queries per month with no credit card required. Enough for most individual developer workflows. Paid plans start at $5/month for higher volume.

Fast Response Times

Brave Search API returns results in under 500ms. The MCP server adds minimal overhead, keeping your conversational flow snappy and responsive.

Structured Results

Results come back as structured data with titles, URLs, descriptions, and metadata — ready for the AI to parse, compare, and synthesize into actionable answers.

Execution Brief

Use this page as a rollout checklist, not just reference text.

Suggest update

Tool Mapping Lens

Organize Tools by Workflow Phase

Catalog-oriented pages work best when users can map discovery, evaluation, and rollout in a clear path instead of reading an undifferentiated list.

Define the job-to-be-done first
Group tools by stage
Prioritize by adoption friction

Actionable Utility Module

Skill Implementation Board

Use this board for Brave Search MCP Server before rollout. Capture inputs, apply one decision rule, execute the checklist, and log outcome.

Input: Objective

Deliver one measurable improvement with brave search mcp server claude code web search review

Input: Baseline Window

20-30 minutes

Input: Fallback Window

8-12 minutes

Decision Trigger	Action	Expected Output
Input: one workflow objective and release owner are defined	Run preview execution with fixed acceptance criteria.	Go or hold decision backed by repeatable evidence.
Input: output quality below baseline or retries increase	Limit scope, isolate root issue, and rerun controlled test.	One confirmed correction path before wider rollout.
Input: checks pass for two consecutive replay windows	Promote to broader traffic with fallback path active.	Stable rollout with low operational surprise.

Execution Steps

Record objective, owner, and stop condition.
Execute one controlled preview run.
Measure quality, latency, and correction burden.
Promote only when pass criteria are stable.

Output Template

tool=brave search mcp server claude code web search review
objective=
preview_result=pass|fail
primary_metric=
next_step=rollout|patch|hold

Share execution feedback

What Is Brave Search MCP Server?

Brave Search MCP Server is a Model Context Protocol integration maintained by Anthropic that connects AI coding assistants to the Brave Search API. It provides two core tools: web search for general internet results and local search for business listings. The server runs via npx and communicates through the MCP stdio protocol.

The fundamental problem it solves is knowledge currency. AI assistants are trained on static data with a cutoff date. When you need to check current library versions, verify API pricing, research new tools, or find recent solutions to technical problems, the assistant cannot access the live web. Brave Search MCP bridges that gap.

For developers, this creates a powerful research loop: ask a question, the agent searches the web, reads results, synthesizes an answer, and can even follow up with additional searches for clarification. You get the depth of web research with the convenience of a conversational interface.

Brave Search was chosen for the official MCP server because of its privacy-first approach (no tracking, no profiling), its independent search index (not proxying Google), and its generous free tier (2,000 queries/month). For most individual developers, the free tier covers all MCP usage without any cost.

How to Calculate Better Results with brave search mcp server claude code web search review

Get a free Brave Search API key at brave.com/search/api. No credit card is required for the free tier (2,000 queries/month).

Install with Claude Code: claude mcp add brave-search -- npx @anthropic/mcp-server-brave-search. Set your BRAVE_API_KEY environment variable.

For Cursor, add the configuration to .cursor/mcp.json. For Codex CLI: codex mcp add brave-search npx "@anthropic/mcp-server-brave-search".

Test by asking your assistant a question about current information: "What is the latest version of Next.js?" or "What is Supabase pricing in 2026?" The agent should return real-time results from the web.

Treat this page as a decision map. Build a shortlist fast, then run a focused second pass for security, ownership, and operational fit.

When a team keeps one shared selection rubric, tool adoption speeds up because evaluators stop debating criteria every time a new option appears.

Worked Examples

Technology comparison research

Ask: "Compare the current pricing of Supabase, PlanetScale, and Neon for a hobby project"
Agent runs 3 separate Brave searches for each service pricing page
Agent synthesizes results into a comparison table with free tier limits, starter pricing, and key differentiators
Ask: "Which one has the best free tier for a side project with auth?"
Agent weighs the search results against your specific requirements and recommends Supabase for its included Auth service

Outcome: A complete technology comparison with current pricing data, produced in under a minute without opening any browser tabs. The agent used live data, not training data, ensuring accuracy.

Finding and applying a recent bug fix

You encounter a cryptic error: "NEXT_REDIRECT in Server Component"
Ask: "Search for this Next.js error and find the latest fix"
Agent searches Brave for the error message, finds recent GitHub issues and Stack Overflow posts
Agent identifies the fix: redirect() must be called outside try/catch blocks in Next.js 15+
Agent explains the solution with code examples based on the search results
You apply the fix and the error resolves

Outcome: A real-time debugging workflow where the agent found a solution from recent web sources that would not exist in its training data. Total time: under 2 minutes.

Frequently Asked Questions

What is the Brave Search MCP Server?