Interactive Feedback

Interactive Feedback MCP enables a human-in-the-loop workflow for AI-assisted development tools by letting you run commands, view their output, and provide textual feedback directly to the AI. It helps you guide the AI to check in with you before completing tasks, reducing unnecessary tool usage and improving collaboration.

How to use

You use this MCP server from your MCP client (for example Cursor or Windsurf) by invoking the interactive_feedback tool whenever you need to ask a question or when you’re ready to complete a task. The AI will run a command, show you the output, and prompt you for feedback. Keep sending feedback until you are satisfied with the result and there is no more input needed.

How to install

Prerequisites: install Python 3.11 or newer and the Python package manager uv.

Install uv as described for your platform.

Get the code and set up the MCP server locally.

Navigate to the project directory, install dependencies with uv sync, and run the server.

Configure the MCP server in your MCP client (for example Cursor) using the provided configuration snippet.

Configuration and example setup

The server is configured to run as a local process via uv. You can specify the directory where the server lives and start it with the appropriate command. The following example shows the exact configuration you would place in your MCP setup.

{
  "mcpServers": {
    "interactive_feedback": {
      "command": "uv",
      "args": [
        "--directory",
        "/Users/fabioferreira/Dev/scripts/interactive-feedback-mcp",
        "run",
        "server.py"
      ],
      "timeout": 600,
      "autoApprove": [
        "interactive_feedback"
      ]
    }
  }
}

Available tools

interactive_feedback

A tool that runs commands, displays their output, and collects user feedback to guide AI-assisted development tasks.