Skip to content

Installation

Get the deepset MCP server running in your preferred environment. The server enables AI agents to interact with the deepset AI platform through standardized tools.

Prerequisites

Before installing the deepset MCP server, ensure you have:

  • A deepset AI platform account
  • An API key for the deepset platform (create one here)
  • Python package manager uv installed

Installing uv

If uv is not installed on your system, install it using one of these methods:

If Python is already installed: 

pipx install uv

Mac/Linux (no Python required): 

curl -LsSf https://astral.sh/uv/install.sh | sh

Windows: 

powershell -c "irm https://astral.sh/uv/install.ps1 | more"

Installation Methods

Choose your installation method based on your use case:

For MCP Clients (Cursor, Claude Desktop)

This method is recommended for users who want to integrate with MCP-compatible AI tools.

The server will be automatically installed and managed when you configure your MCP client:

Cursor Configuration:

Create a file named mcp.json in your .cursor directory at the project root:

{
  "mcpServers": {
    "deepset": {
      "command": "uvx",
      "args": ["deepset-mcp"],
      "env": {
        "DEEPSET_WORKSPACE": "your_workspace_name",
        "DEEPSET_API_KEY": "your_api_key"
      }
    }
  }
}

💡 Note: The DEEPSET_WORKSPACE is optional. If not provided, the AI agent will need to specify the workspace name when calling tools.

Claude Desktop Configuration:

Edit your Claude Desktop configuration file at /Users/your_user/Library/Application Support/Claude/claude_desktop_config.json (Mac):

{
  "mcpServers": {
    "deepset": {
      "command": "uvx",
      "args": ["deepset-mcp"],
      "env": {
        "DEEPSET_WORKSPACE": "your_workspace_name",
        "DEEPSET_API_KEY": "your_api_key"
      }
    }
  }
}

💡 Note: The DEEPSET_WORKSPACE is optional. If not provided, the AI agent will need to specify the workspace name when calling tools.

For Direct Python Integration

Install the package directly for use in your Python applications:

Using pip: 

pip install deepset-mcp

Using uv: 

uv add deepset-mcp

After installation, you can use the server programmatically in your Python code:

from deepset_mcp import configure_mcp_server
from mcp.server.fastmcp import FastMCP

# Create and configure the MCP server
mcp = FastMCP("Deepset MCP Server")
configure_mcp_server(mcp_server_instance=mcp)

# Run the server
mcp.run()

For Command Line Usage

Run the server directly from the command line:

Using uvx (recommended): 

uvx deepset-mcp --workspace your_workspace --api-key your_api_key

If installed globally: 

deepset-mcp --workspace your_workspace --api-key your_api_key

💡 Note: The --workspace parameter is optional. If omitted, the AI agent will need to specify the workspace name when calling tools.

Using Docker

For containerized environments, use the official Docker image:

{
  "mcpServers": {
    "deepset": {
      "command": "/usr/local/bin/docker",
      "args": [
        "run",
        "-i",
        "-e",
        "DEEPSET_WORKSPACE",
        "-e",
        "DEEPSET_API_KEY",
        "deepset/deepset-mcp-server:main"
      ],
      "env": {
       "DEEPSET_WORKSPACE": "your_workspace_name",
       "DEEPSET_API_KEY": "your_api_key"
     }
    }
  }
}

💡 Note: The DEEPSET_WORKSPACE environment variable is optional. If not provided, the AI agent will need to specify the workspace name when calling tools.

Verification

After installation, verify the server is working:

  1. For MCP clients: Check that the deepset server appears in your client's tools menu
  2. For command line: Run uvx deepset-mcp --list-tools to see available tools
  3. For Python integration: Import the module without errors

Next Steps