Gemini Data

You can analyze datasets, generate research content, and distribute results with a single MCP server that integrates Gemini’s Gemini Flash 2 AI. This server automates data analysis, crafts professional research outputs, and delivers results by email, all through convenient MCP client interactions.

How to use

You interact with the Gemini Data Analysis MCP Server through an MCP client to perform three core actions: analyze data, generate thinking content, and send emails with AI-generated subject lines and content. Use analyze-data to process Excel or CSV files and produce interactive HTML reports with charts. Use generate-thinking to create research papers, technical documents, and business intelligence content. Use send-email to deliver AI-generated outputs by email, with HTML or plain text content, optional images, and tracking.

How to install

Prerequisites you need before installing: Node.js (v16 or higher), TypeScript, Claude Desktop, Google Gemini API Key, and an SMTP email account for sending emails.

Install steps you should follow exactly in order:

git clone falahgs/gemini-data-analysis-research-mcp-server
cd gemini-data-analysis-research-mcp-server
npm install

Create an environment file with your API key and email credentials. The server reads GEMINI_API_KEY for Gemini access and NODEMAILER_* for sending emails.

GEMINI_API_KEY=your_api_key_here
NODEMAILER_EMAIL=your.email@gmail.com
NODEMAILER_PASSWORD=your_app_password_here

Build and start the server. This compiles TypeScript to JavaScript and launches the MCP server runtime.

npm run build
npm run start

Configure Claude Desktop to connect to the MCP server by editing the Claude configuration file to include the runtime command and environment variables as shown.

{
  "mcpServers": {
    "Gemini Data Analysis": {
      "command": "node",
      "args": ["path/to/gemini-data-analysis-email-generator/dist/index.js"],
      "cwd": "path/to/gemini-data-analysis-email-generator",
      "env": {
        "GEMINI_API_KEY": "your_api_key_here",
        "NODEMAILER_EMAIL": "your.email@gmail.com",
        "NODEMAILER_PASSWORD": "your_app_password_here"
      }
    }
  }
}

Additional configuration notes

The server exposes three MCP tools that you can invoke through your MCP client: analyze-data, generate-thinking, and send-email. Each tool accepts a structured arguments payload and returns results to the configured output directories.

Security and best practices

Store API keys and credentials securely. Do not share your .env file. For Gmail, use an app password rather than your main account password. Be cautious with email content and avoid including sensitive information in examples.

Troubleshooting

If you encounter issues, verify the environment variables are loaded, ensure the API key is valid, confirm Claude Desktop is configured with the correct path to the compiled index.js, and check email credentials. If email delivery fails, confirm SMTP settings and that the app password is active.

Examples of common usage patterns

Data analysis: prepare a base64-encoded file payload, specify the file name, choose a detailed analysis type, and set an output directory. The server returns an HTML report with interactive charts and a textual analysis.

API reference (high level)

The server provides three primary operations: analyze-data for data analysis with AI insights, generate-thinking for AI-generated research content, and send-email for composing and delivering emails with optional images and HTML content.

Available tools

analyze-data

Analyze Excel or CSV data with automatic parsing, statistical insights, interactive charts, AI-powered notes, HTML reports, and optional email delivery.

send-email

Generate AI-driven email content and deliver via SMTP with HTML/plain text, optional images, subject line generation, and delivery tracking.

generate-thinking

Create AI-generated research content such as papers, technical docs, data summaries, and BI reports with optional email delivery.