home / mcp / mcp interactive service mcp server

MCP Interactive Service MCP Server

Provides interactive user-aided AI collaboration with option selection and information requests across CLI, PyQt, and Web interfaces.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "danielzhao1990-interaction-mcp": {
      "url": "http://127.0.0.1:8000/sse"
    }
  }
}

MCP Interactive Service enables high-frequency, interactive communication between AI tools and users. It lets AI models confirm decisions, request additional information, and let you steer multi-step tasks across multiple interfaces, reducing wasted calls and keeping the AI aligned with your goals.

How to use

You use the MCP server by connecting your AI tool to it via one of the supported transport methods and then invoking the interactive tools when input or confirmation is needed. You can drive option selections, request supplementary information, or prompt the AI to confirm critical decisions before proceeding. Use the UI that best fits your workflow: a CLI-based dialog, a PyQt desktop dialog, or a web-based interface, depending on your environment and preferences.

How to install

Prerequisites: make sure you have Python installed on your system. If you prefer using a pre-built executable on Windows, you can skip Python installation once you download the executable.

Option A. Use a pre-compiled executable (Recommended for Windows) Download the latest pre-compiled executable from the distribution page. Run the executable directly to start the MCP Interactive Service.

Option B. Install from source (Python) Follow these steps to set up the environment and run from source.

Step 1: Install base dependencies using pip.

pip install -r requirements/requirements-base.txt

Step 2: If you want a specific UI, install the corresponding dependencies.

pip install -r requirements/requirements-pyqt.txt # PyQt UI

pip install -r requirements/requirements-web.txt # Web UI

Step 3: Start the program with your preferred UI.

python main.py run --ui=cli # Command line interface (default)

python main.py run --ui=web # Web interface

python main.py run --ui=pyqt # PyQt interface

Step 4: If you need to customize host/port or log level, adjust startup options.

Configuration and usage details

Connect to MCP using either stdio (local, direct process) or SSE (networked) transport. The stdio method runs the MCP server as a local process and communicates via standard input/output. The SSE method talks over HTTP to a live server. Use the following configurations to connect from your AI tool.

"ai-interaction": {
  "command": "python",
  "args": ["path/to/main.py", "run", "--transport", "stdio", "--ui", "cli"],
  "env": {}
}

Tools and how they work

The MCP service exposes tools to manage interactive AI dialogs. You will typically use these when the AI needs a user choice or extra information.

Option Selection lets the user pick from a list of options or provide a custom answer.

Information Supplement asks the user to provide missing details or context.

Examples of common configurations and flows

Two common ways to run MCP locally:

1) Python stdio-based startup (local process) with CLI UI

2) Executable-based startup (Windows) with PyQt UI

Security and best practices

Only expose the MCP service to trusted clients and networks. Use strong access controls for the HTTP interface, and prefer local stdio transport for machine-local integrations to minimize exposure.

Available tools

select_option

Presents a list of options to the user and returns the chosen index or a custom input, enabling precise user-directed selection.

request_additional_info

Requests supplementary information from the user when the AI model needs more complete context or data to proceed.