Brave Search MCP/SSE Server
@Shoofio
Stream Brave Search (web & local) results via a Model Context Protocol (MCP) / Server-Sent Events (SSE) interface. Acts as a centralized, observable tool provider for AI models, deployable via Docker or Helm.
Overview
What is Brave Search MCP/SSE Server?
An implementation of the Model Context Protocol (MCP) using Server-Sent Events (SSE) that integrates the Brave Search API, providing AI models and other clients with web and local search capabilities through a streaming interface.
How to use Brave Search MCP/SSE Server?
Pull the Docker image from Docker Hub (shoofio/brave-search-mcp-sse) or deploy via Helm chart from https://shoofio.github.io/brave-search-mcp-sse/. Alternatively, run locally after cloning, building, and configuring the BRAVE_API_KEY environment variable. Clients connect via HTTP GET to the SSE endpoint (e.g., http://localhost:8080/sse).
Key Features of Brave Search MCP/SSE Server
- Web search via Brave's independent index.
- Local search with smart fallback to web search.
- Real‑time streaming results via SSE.
- Centralized API key management.
- Robust logging and observability.
- Docker and Helm deployment support.
Use Cases of Brave Search MCP/SSE Server
- Provide AI assistants with real‑time web search results.
- Enable local business discovery for location‑aware chatbots.
- Centralize Brave Search access for multiple internal clients.
- Integrate search capabilities into MCP‑compatible IDEs (e.g., Cursor).
FAQ from Brave Search MCP/SSE Server
How does this compare to using the Brave Search API directly?
This server wraps the API into an MCP/SSE interface, enabling seamless integration with AI models and clients that support the Model Context Protocol.
What are the runtime dependencies?
Node.js v22+ (for local development), Docker (for container deployment), or kubectl/Helm (for Kubernetes). A valid Brave Search API key is always required.
Where does the search data come from and is it stored?
All search results are fetched live from the Brave Search API. The server does not store any search data locally.
What transport does the server use?
It uses Server‑Sent Events (SSE) over HTTP, adhering to the Model Context Protocol (MCP). Clients connect to the /sse endpoint.
Are there any usage limits built in?
Web search supports count (1–20, default 10) and offset (0–9). Local search supports count (1–20, default 5). Rate limiting is logged but not enforced by the server.
