Brave Search
Web and local search results via the Brave Search API — no scraping search engines directly.
npx -y @modelcontextprotocol/server-brave-searchRather than pointing a browser-automation server at a search engine’s results page and scraping it, the Brave Search server calls Brave’s actual Search API and returns structured results — faster, more reliable, and without the fragility of parsing a page that can change its markup at any time. It supports both general web search and local/business search.
You’ll need a BRAVE_API_KEY, obtained from the Brave Search API dashboard; the free tier covers 2,000 queries a month, which is enough for most agent workflows during development and light production use before you need to consider a paid plan.
Like PostgreSQL, this server has been moved from Anthropic’s actively maintained reference-servers repository into servers-archived. It still works exactly as documented, but we mark it Community verification to be transparent that it’s no longer receiving official updates — worth knowing if you’re choosing between this and a similarly-scoped, actively maintained alternative for a production deployment.
Similar MCP Servers
Context7
Pulls current, version-accurate library documentation directly into context — no stale training-data APIs.
Fetch
Fetches a URL and converts it to clean Markdown so agents can read the live web.
Playwright
Fast, deterministic browser automation via accessibility snapshots — no screenshots required.