Brave-Gemini-Research

The Brave-Gemini Research MCP Server exposes three primary tools to empower AI assistants with powerful web, local, and academic analysis capabilities. The Web Search Tool lets you query the Brave Search API to retrieve general internet results, the Local Search Tool helps locate businesses and places of interest, and the Research Paper Analysis Tool uses Google's Gemini model to analyze academic papers. These tools are exposed through a standardized MCP interface, enabling seamless use by assistants like Claude Desktop for tasks such as web research, location discovery, and in-depth paper analysis. To utilize the tools, send structured tool requests via the MCP protocol to the running server; responses will include curated results, summarized findings, and, when applicable, structured analysis from the Gemini model.

In practical use, you can prompt Claude or another assistant to perform: (1) a web search for the latest AI research papers, (2) a local search for coffee shops near a specified location, and (3) a comprehensive analysis of a research paper by providing the full text content to the gemini_research_paper_analysis function with your chosen analysis type (e.g., comprehensive, summary, literature review). The server coordinates requests to Brave, local search services, and Gemini to return cohesive results within your conversation.

Prerequisites:

• Node.js v18+ installed
• Brave Search API key (BRAVE_API_KEY)
• Google API key for Gemini integration (GOOGLE_API_KEY)
• Optional: Claude Desktop for integrated workflow

1. Clone the repository:

   git clone https://github.com/falahgs/brave-gemini-research-mcp.git
   cd brave-gemini-research-mcp
   

2. Install dependencies:

   npm install
   

3. Create a .env file with your API keys (or set environment variables at runtime):

   BRAVE_API_KEY=your_brave_api_key
   GOOGLE_API_KEY=your_google_api_key
   

4. Build the project (TypeScript to JavaScript):

   npm run build
   # or manually
   npx tsc
   

5. Run the server (example):

   • Ensure environment variables are set, then start the server:

     BRAVE_API_KEY=your_brave_api_key GOOGLE_API_KEY=your_google_api_key node dist/index.js
     

6. If using Claude Desktop integration, configure the MCP server entry with the appropriate path to dist/index.js and environment variables as shown in the Claude integration instructions.

Tips and common issues:\n- Ensure your API keys are correctly set in the environment before starting the server.\n- If you modify TypeScript sources, re-run npm run build or npx tsc to regenerate dist/index.js.\n- In Claude Desktop, use absolute paths for the server executable and working directory when configuring the MCP entry.\n- If you encounter timeout issues, increase timeoutMs in the Claude Desktop configuration and verify that BRAVE_API_KEY and GOOGLE_API_KEY have the required permissions.\n- For Gemini analysis, ensure the Google API key has access to Gemini models and that the paper content complies with token limits.\n- Logs can be verbose under DEBUG=mcp:*, helpful for debugging integration problems.