deep-research-mcp-server

Overview

The Deep Research MCP Server is a TypeScript-based implementation that enables AI-powered research using Google Gemini 2.5 Flash with Google Search Grounding. It conducts iterative deep research with query refinement and result analysis, carrying forward learned context. The server features structured JSON outputs, semantic splitting for robust summarization, batching with caching for performance, and generates professional Markdown reports with abstract, table of contents, methodology, and references. It's designed to be both a standalone CLI tool and an MCP server for seamless agent integration.

Try asking AI

After installing, here are 6 things you can ask your AI assistant:

• deep-researchquery: string, depth?: number, breadth?: number, existingLearnings?: string[]

  Conduct iterative, deep research using Google Gemini 2.5 Flash with search grounding and URL context.

  Try:What are the latest developments in AI research agents?Try:Explain the principles of blockchain technologyTry:State of multi-agent research agents in 2025

Comparable tools

Installation

git clone https://github.com/ssdeanx/deep-research-mcp-server
cd deep-research-mcp-server

npm install

GEMINI_API_KEY="your_gemini_key"
GEMINI_MODEL=gemini-2.5-flash
CONCURRENCY_LIMIT=5

npm run build

node --env-file .env.local dist/mcp-server.js

{
  "mcpServers": {
    "deep-research": {
      "command": "node",
      "args": ["--env-file", ".env.local", "dist/mcp-server.js"]
    }
  }
}

FAQ

What models does this MCP server support?

The server is built around Google Gemini 2.5 Flash, with configurable model parameters via environment variables.

Does it require web scraping tools?

No, it uses Google Search Grounding via Gemini's native tools, eliminating the need for web scraping dependencies.

Can I control the depth and breadth of research?

Yes, the server accepts depth and breadth parameters (1-5 scale) to precisely control the exploration scope.

Compare deep-research-mcp-server with