Brave Search MCP Server

The Brave Search MCP Server bridges Claude and other Model Context Protocol clients to the Brave Search API, delivering both web‑wide and local business search capabilities. It addresses a common developer pain point: integrating a privacy‑focused, ad‑free search engine into conversational agents without writing custom HTTP wrappers or managing proxy settings. By exposing a pair of high‑level tools— and —the server lets AI assistants ask for up‑to‑date information, news, or nearby services with a single API call, while handling pagination, safety filtering, and automatic fallbacks internally.

At its core, the server translates MCP tool calls into Brave API requests. For web search, developers can specify a query string, the number of results per page (up to 20), and an offset for pagination. Local search focuses on businesses, restaurants, or services in a geographic area; if no local hits are found, the server transparently switches to a web search so the user never encounters an empty response. This dual‑mode approach is especially valuable in location‑aware applications where a user might first seek nearby recommendations and then broaden the scope to general information.

Key features include:

• Flexible filtering – control result types, safety levels, and content freshness to match the tone and policy requirements of different applications.
• Proxy support – automatically route requests through / or a custom , simplifying deployment in restricted networks.
• Pagination and freshness controls – retrieve subsequent pages of results or limit to the most recent content, enabling conversational agents to surface the latest news or updates.

Typical use cases span from travel assistants that recommend restaurants and attractions, to customer support bots that fetch the latest product information or news articles. In a developer workflow, an MCP client can simply invoke or as part of a prompt, receive structured JSON results, and embed them in the assistant’s reply—all without handling API keys or HTTP details. The server’s automatic fallback logic and proxy configuration further reduce boilerplate, making it a standout choice for privacy‑conscious teams looking to enrich their AI applications with reliable search data.