brave_web_searchLABSCO SUMMARY
Brave's own index as tools — web, news, image, video, place and LLM-grounding results behind one key.
Full local search and extra_snippets need a Pro plan, and local search quietly falls back to web search without one. The HTTP transport binds to 127.0.0.1 and carries no authentication, so widening BRAVE_MCP_HOST belongs on a trusted network only.
What it is
An MCP server published by Brave that exposes the Brave Search API as agent tools. Queries run against Brave's own web index and come back as structured results, so an agent gets ranked answers without fetching and parsing pages itself.
What you get
- Web results filtered by country, language, freshness and safesearch, with custom re-ranking through Goggles definitions.
- News results with breaking-news indicators, defaulting to a freshness window of the last 24 hours.
- Image results with dimensions and confidence scores, and video results with thumbnail metadata.
- Local business and place lookup by coordinates and radius or by location string, returning addresses, opening hours, contact info, ratings, photos, categories and timezone.
- Pre-extracted page content for LLM grounding and RAG, bounded by a token budget you set per request and per URL.
- AI summaries from Brave's summarization API, with optional entity information and inline source references.
Requirements
A Brave Search API account and a key, supplied as BRAVE_API_KEY, or as a path to a key file in BRAVE_API_KEY_FILE, which takes precedence when both are set. Node.js 22.x or higher.
Setup effort
Add the configuration to your client
USE CASES
01
Ground an answer in current web results
Run a query and read back ranked results with snippets and URLs, narrowed by country, language and a time window.
02
Fill a RAG context window
Pull pre-extracted page text under an explicit token cap, with limits on how many URLs and snippets come back and how much each URL contributes.
03
Summarize what the web says about a topic
Search with summary generation enabled, then pass the returned summary key back to get written prose with source references.
04
Track a story as it develops
Query news with a freshness filter of past day, week, month, year or an explicit date range.
05
Find businesses in an area
Search points of interest around a coordinate pair and radius, or around a location string, and read hours, contact details and ratings.
06
Collect media for a brief
Gather images and videos on a topic with their source URLs and metadata, under a safesearch level you choose.
8 tools
Listed in project docs. Your client may show a different list.
brave_local_searchbrave_video_searchbrave_image_searchbrave_news_searchbrave_summarizerbrave_place_searchbrave_llm_contextBEFORE YOU INSTALL
What to know first
Tools you get
8 tools named in the project's documentation.
Requires
BRAVE_API_KEY in your environment
Where it runs
All 3 documented setups run the server on your machine.
INSTALLATION
Connect your client
MCP Clients
Claude Desktop — Docker
CLAUDE DESKTOP — DOCKER · JSON
{
"mcpServers": {
"brave-search": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"-e",
"BRAVE_API_KEY",
"docker.io/mcp/brave-search"
],
"env": {
"BRAVE_API_KEY": "<YOUR_BRAVE_API_KEY>"
}
}
}
}VS Code — npx
VS CODE — NPX · JSON
{
"inputs": [
{
"password": true,
"id": "brave-api-key",
"type": "promptString",
"description": "Brave Search API Key"
}
],
"servers": {
"brave-search-mcp-server": {
"command": "npx",
"args": [
"-y",
"@brave/brave-search-mcp-server",
"--transport",
"stdio"
],
"env": {
"BRAVE_API_KEY": "${input:brave-api-key}"
}
}
}
}VS Code — Docker
VS CODE — DOCKER · JSON
{
"inputs": [
{
"password": true,
"id": "brave-api-key",
"type": "promptString",
"description": "Brave Search API Key"
}
],
"servers": {
"brave-search": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"-e",
"BRAVE_API_KEY",
"mcp/brave-search"
],
"env": {
"BRAVE_API_KEY": "${input:brave-api-key}"
}
}
}
}
