# Interactive Feedback MCP for macOS A native macOS MCP server for collecting interactive user feedback with AppleScript dialogs and image support. ![image](https://github.com/user-attachments/assets/7d321b56-303c-4045-9ec5-de865c1c5b11) ![image](https://github.com/user-attachments/assets/a0652fa6-5ddb-48c5-a3cb-067a95748110) ## ✨ What's Different This is a macOS-native reimplementation of the original [interactive-feedback-mcp](https://github.com/noopstudios/interactive-feedback-mcp) by Fábio Ferreira ([@fabiomlferreira](https://x.com/fabiomlferreira)) with the following enhancements: - **Native macOS Dialogs**: Uses AppleScript instead of web UI - **Image Support**: Built-in image picker and screenshot capture - **Lightweight**: No web server dependencies (Express, CORS, etc.) - **Better Integration**: Seamless macOS user experience ## 🛠 Installation 1. Clone and install: ```bash git clone https://github.com/gmh5225/interactive-feedback-macos-mcp.git cd interactive-feedback-macos-mcp npm install ``` 2. Add to your Cursor MCP configuration: ```json { "mcpServers": { "interactive-feedback-macos-mcp": { "command": "node", "args": ["/path/to/interactive-feedback-macos-mcp/src/mcp-server-macos.js"], "autoApprove": ["collect_feedback", "pick_image", "get_image_info", "take_screenshot"] } } } ``` ## 🔧 Available Tools - **`collect_feedback`**: Native dialog for user feedback with optional image attachment - **`pick_image`**: macOS file picker for image selection - **`get_image_info`**: Get image metadata and base64 content - **`take_screenshot`**: Capture area or full screen screenshots ## 📝 Usage Add this to your AI assistant prompt for best results: > Whenever you want to ask a question, always call the MCP `interactive-feedback-macos-mcp.collect_feedback`. > Whenever you're about to complete a user request, call the MCP `interactive-feedback-macos-mcp.collect_feedback` instead of simply ending the process. ## License MIT License ## 🙏 Credits Based on [interactive-feedback-mcp](https://github.com/noopstudios/interactive-feedback-mcp) by Fábio Ferreira ([@fabiomlferreira](https://x.com/fabiomlferreira)).