home / mcp / mcp multi-agent deep researcher mcp server

MCP Multi-Agent Deep Researcher MCP Server

Provides a multi-agent research workflow with web search, analysis, and writing via a local MCP server.

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "anubhav-77-dev-mcp-multi-agent-deep-researcher": {
      "command": "poetry",
      "args": [
        "run",
        "python",
        "Multi-Agent-deep-researcher-mcp-windows-linux/server.py"
      ],
      "env": {
        "MODEL_NAME": "phi3:latest",
        "LINKUP_API_KEY": "your_linkup_api_key_here",
        "OPENAI_API_KEY": "YOUR_OPENAI_API_KEY",
        "OLLAMA_BASE_URL": "http://localhost:11434"
      }
    }
  }
}

You will run a hosted MCP server that coordinates multiple AI agents to search the web, analyze findings, and generate clear, structured written outputs. This server uses the Model Context Protocol (MCP) to integrate with MCP clients, enabling you to perform quick searches or full research workflows through a single, cohesive interface.

How to use

You connect to the MCP server from a compatible MCP client. The standard workflow lets three specialized agents collaborate: a Web Searcher to gather information, a Research Analyst to synthesize and verify findings, and a Technical Writer to produce clear, structured output. You can send a query for a quick search or request a full research report. The system returns sources, insights, and a well-formatted final document that you can copy or export.

How to install

Prerequisites: you need Python and Poetry, and you should have access to a LinkUp API key to enable web search. Ensure Ollama is installed for local AI processing if you want to run the phi3 model locally.

# Step 1: Clone the project
git clone https://github.com/anubhav-77-dev/MCP-Multi-Agent-Deep-Researcher.git
cd MCP-Multi-Agent-Deep-Researcher

# Step 2: Set up the environment and dependencies
python3 setup.py

# Step 3: Launch the application
python3 launcher.py

Configuration and MCP client integration

To integrate this MCP server with an MCP client, use the following configuration. This config starts the local MCP server process and provides the required environment variable for the LinkUp API key.

{
  "mcpServers": {
    "crew_research": {
      "command": "poetry",
      "args": ["run", "python", "Multi-Agent-deep-researcher-mcp-windows-linux/server.py"],
      "env": {
        "LINKUP_API_KEY": "your_linkup_api_key_here"
      }
    }
  }
}

Usage notes

The MCP server exposes a REST-like API for quick searches and full research, and also provides a web interface for interactive use. You can run a quick search to get immediate results or trigger a full multi-agent analysis to obtain a comprehensive report. Ensure your environment variables are set correctly, especially the LinkUp API key and the local Ollama setup if you plan to run the local phi3 model.

Troubleshooting tips

If you encounter issues, verify that all services are running and accessible. Check that the Ollama service is up when using local AI processing, and confirm that the LinkUp API key is valid and has not expired. Review logs produced by the launcher for any errors and restart services as needed.

Available tools

LinkUp Search Tool

Web search integration using the LinkUp API to retrieve relevant sources and provide a structured summary of results with sources.

Ollama Phi3 Tool

Local AI processing using the phi3 model served by Ollama for offline inference.

CrewAI Orchestrator

Orchestrates the three agents (Web Searcher, Research Analyst, Technical Writer) to perform coordinated workflows.