AI Interaction Tool

# AI Interaction Tool - MCP Server **Modern AI interaction tool with advanced UI and powerful features for Model Context Protocol (MCP)** ## 🚀 Core Features ### 🎯 Main Capabilities - **Interactive UI Popup** for content input and conversation control - **File/Folder Attachment** from workspace with validation and preview - **🖼️ Image Attachment System** with drag & drop, multi-image support - **Multi-language Support** (English/Vietnamese) - **Maximum Cognitive Power** activation for peak AI performance - **Tag-based Output Format** integrated with system prompt rules - **Workspace-aware Path Processing** for cross-project compatibility ### 🔧 New in v2.2.0 (Latest) - **🖼️ Image Attachment Support** with drag & drop functionality - **🛡️ Security Enhanced** - secure path storage in user_images directory - **💾 Persistent Image State** - checkbox state saves correctly - **🎯 Multi-image Management** - attach, preview, and remove multiple images - **🔄 Database Auto-cleanup** - automatic image cleanup when disabled ### 🔧 Previous v2.1.0 - **Enhanced UI/UX** with modern PyQt5 interface - **Structured Tag-based Output** for perfect AI agent integration - **Debounce Configuration** with smart auto-save mechanisms - **Cursor IDE Integration** with comprehensive setup guide ## 📋 Installation & Setup Guide ### 📥 Step 1: Clone Repository ```bash git clone https://github.com/your-username/AI-interaction.git cd AI-interaction ``` ### 🐍 Step 2: Install Python - **Requirement**: Python 3.8+ - Download from [python.org](https://www.python.org/downloads/) - Or use package manager: ```bash # Windows with Chocolatey choco install python # macOS with Homebrew brew install python # Ubuntu/Debian sudo apt update && sudo apt install python3 python3-pip ``` ### 📦 Step 3: Install Dependencies ```bash # Using pip pip install -r requirements.txt # Or using uv (recommended for performance) pip install uv uv pip install -r requirements.txt ``` ### ⚙️ Step 4: Configure MCP Server in Claude Desktop Add the following configuration to Claude Desktop config file: **Config file paths:** - **Windows**: `%APPDATA%\Claude\claude_desktop_config.json` - **macOS**: `~/Library/Application Support/Claude/claude_desktop_config.json` - **Linux**: `~/.config/claude/claude_desktop_config.json` **Configuration content:** ```json { "mcpServers": { "AI_interaction": { "command": "python", "args": ["E:/MCP-servers-github/AI-interaction/mcp_server.py"], "stdio": true, "enabled": true } } } ``` **⚠️ Important**: Replace `E:/MCP-servers-github/AI-interaction/mcp_server.py` with the absolute path to `mcp_server.py` on your system. ### 🧠 Step 5: Configure AI Agent Rules (REQUIRED) **For proper AI agent operation with ai_interaction tool, you MUST setup custom instructions:** #### 📋 **How to Add Custom Instructions:** 1. **Open Claude Desktop** or access Claude web interface 2. **Find "Custom Instructions"** or "Add custom instructions" in settings 3. **Copy entire content** from one of the rule files: - **🇻🇳 Vietnamese**: `rule_for_ai_VI.txt` - **🇺🇸 English**: `rule_for_ai_EN.txt` 4. **Paste into custom instructions** field and save #### 🎯 **Why This is Necessary:** - ✅ **Behavioral Framework**: Rules define how AI agent processes ai_interaction output - ✅ **Thinking Protocols**: Activates high-level thinking patterns for quality responses - ✅ **Ultra-Enhancement Modes**: 10 cognitive modes for maximum performance - ✅ **Tag Processing**: Reads and processes control tags like `<AI_INTERACTION_CONTINUE_CHAT>` - ✅ **Continue Logic**: Auto-recall ai_interaction when `continue_chat=true` #### 📁 **Rule Files Location:** ``` AI-interaction/ ├── rule_for_ai_VI.txt # Vietnamese rules ├── rule_for_ai_EN.txt # English rules └── ... ``` #### ⚡ **Quick Setup Commands:** ```bash # View Vietnamese rules content cat rule_for_ai_VI.txt # View English rules content cat rule_for_ai_EN.txt # Copy to clipboard (Windows) type rule_for_ai_VI.txt | clip # Copy to clipboard (macOS) cat rule_for_ai_VI.txt | pbcopy # Copy to clipboard (Linux) cat rule_for_ai_VI.txt | xclip -selection clipboard ``` ### 🚀 Step 6: Configure Cursor IDE (Recommended) **Cursor is the recommended IDE for AI development with this tool:** #### 📋 **Cursor Setup Steps:** 1. **Download Cursor**: [https://cursor.sh/](https://cursor.sh/) 2. **Install and open workspace**: Open AI-interaction folder 3. **Configure MCP in Cursor**: - Open Command Palette (`Cmd/Ctrl + Shift + P`) - Search "Configure MCP Servers" - Add AI_interaction server config 4. **Setup custom instructions**: - Copy content from `rule_for_ai_VI.txt` or `rule_for_ai_EN.txt` - Paste into "Custom Instructions" field in custom mode Agent:    #### 🎯 **Cursor Advantages:** - ✅ **Native MCP Support**: Built-in integration with MCP servers - ✅ **AI-First IDE**: Optimized for AI development workflows - ✅ **Real-time Suggestions**: Context-aware code completion - ✅ **Advanced Debugging**: Enhanced debugging for MCP tools - ✅ **Performance**: Faster than traditional IDEs for AI projects ### 🚀 Step 7: Launch and Test !!! -----> In your terminal: python E:\MCP-servers-github\AI-interaction\main.py --ui **⚠️ Important**: Replace `E:/MCP-servers-github/AI-interaction/mcp_server.py` with the absolute path to `mcp_server.py` on your system. ---> AUTO SHOW UI: <img width="1164" height="930" alt="image" src="https://github.com/user-attachments/assets/b2b633d1-bc62-4c66-ad21-1c2b8eb71eb5" /> 1. **Restart Claude Desktop/Cursor** after configuring MCP server 2. **Test connection** by calling `ai_interaction` tool 3. **Test UI popup** to verify functionality 4. **Validate rule integration** through AI agent responses ## 📦 Package Structure ``` AI-interaction/ ├── ai_interaction_tool/ # Main interaction tool package │ ├── core/ # Core dialog and configuration │ │ ├── dialog.py # InputDialog with PyQt5 UI │ │ └── config.py # Configuration management │ ├── ui/ # Interface and styling │ │ ├── file_dialog.py # File attachment dialogs │ │ ├── file_tree.py # File system tree view │ │ ├── image_attachment.py # 🖼️ Image attachment with drag & drop │ │ └── styles.py # Modern UI styling │ ├── utils/ # Utilities and multi-language │ │ ├── translations.py # Multi-language support │ │ └── file_utils.py # File operation utilities │ ├── engine.py # Main entry point │ ├── description.py # Detailed tool description │ └── __init__.py # Package exports ├── user_images/ # 🛡️ Secure image storage directory ├── main.py # Legacy entry point ├── mcp_server.py # MCP server implementation ├── requirements.txt # Python dependencies ├── pyproject.toml # Project configuration └── README.md # This file ``` ## 🎮 Usage Guide ### Available Tools in MCP Server #### 1. **ai_interaction**: Main Interactive Tool - **Function**: Creates UI popup for user input with file/image attachment - **Output**: Structured tag-based format with image support - **Integration**: Perfect integration with system prompt rules - **Use cases**: - Input complex content with formatting - Attach files/folders from workspace - **🖼️ Attach images with drag & drop functionality** - **📷 Multi-image support with preview and management** - Control AI thinking modes and reasoning levels ### Basic Usage Examples ```python # Programmatic usage from ai_interaction_tool import ai_interaction # Launch interactive interface result = ai_interaction() print(result) # Structured output with tags ``` ### 🖼️ Image Attachment Features #### 📷 **Core Image Capabilities** - **Drag & Drop Support**: Drag images directly into the UI - **Multi-image Management**: Attach, preview, and remove multiple images - **Format Support**: PNG, JPG, JPEG, GIF, BMP, WEBP - **Secure Storage**: Images stored safely in `user_images/` directory - **Base64 Encoding**: Automatic conversion for AI processing - **Preview System**: Click images to view larger versions - **Persistent State**: Save images option with checkbox persistence #### 🎯 **How to Use Image Attachment** 1. **Attach Button**: Click "📷 Attach Images" to select files 2. **Drag & Drop**: Drag images from file explorer directly to UI 3. **Paste Support**: Paste images from clipboard (Ctrl+V) 4. **Multiple Images**: Attach as many images as needed 5. **Remove Images**: Click X button on individual image previews 6. **Clear All**: Use "🗑️ Clear Images" to remove all at once 7. **Save Toggle**: Check/uncheck "Save images" to control persistence #### 🛡️ **Security & Privacy** - **Local Only**: All images stored locally in `user_images/` - **No External Access**: No uploads or external connections - **Relative Paths**: Only relative paths stored in config for security - **User Control**: Users control what images to attach and save - **Auto-cleanup**: Images automatically cleaned when save disabled ### Output Format AI Interaction Tool uses clean tag-based format: ``` User message content with natural line breaks <AI_INTERACTION_ATTACHED_FILES> FOLDERS: - workspace_name/relative/path/to/folder FILES: - workspace_name/relative/path/to/file.js </AI_INTERACTION_ATTACHED_FILES> <AI_INTERACTION_WORKSPACE>workspace_name</AI_INTERACTION_WORKSPACE> <AI_INTERACTION_CONTINUE_CHAT>true/false</AI_INTERACTION_CONTINUE_CHAT> ``` **Note**: When images are attached, they are automatically converted to base64 format and included in the response for AI processing. ## 🔧 Troubleshooting ### Common Issues 1. **"Command not found" error** - Check Python is installed and in PATH - Verify absolute path in MCP config 2. **"Module not found" error** - Run `pip install -r requirements.txt` - Check virtual environment if using one 3. **UI not displaying** - Ensure PyQt5 is installed correctly - Check display settings and desktop environment 4. **File attachment not working** - Verify file permissions and access rights - Check workspace path configuration 5. **🖼️ Image attachment issues** - Ensure PyQt5 is properly installed for image processing - Check `user_images/` directory permissions - Verify image formats: PNG, JPG, JPEG, GIF, BMP, WEBP supported - Clear config if images not loading: Remove `last_attached_images` from config.json 6. **MCP Connection Issues in Cursor** - Verify MCP server configuration in Cursor settings - Check process running with `ps aux | grep mcp_server` - Restart Cursor after config changes ### Debug Mode To debug issues, run server directly: ```bash python mcp_server.py ``` For Cursor debugging: ```bash # Check MCP server logs in Cursor # Open Developer Tools → Console # Look for MCP connection messages ``` ## 🔄 Version History - **v2.2.0** (Latest): 🖼️ **Image Attachment System** - Complete image support with drag & drop, multi-image management, security enhancements, and persistent state - **v2.1.0**: Enhanced UI/UX, Cursor IDE integration, Debounce config system - **v2.0.0**: Refactored architecture with modern PyQt5 UI - **v1.x**: Core functionality and basic features ### 🎯 **v2.2.0 Detailed Changes:** - ✅ **Image Attachment UI**: Full drag & drop interface with preview system - ✅ **Multi-format Support**: PNG, JPG, JPEG, GIF, BMP, WEBP compatibility - ✅ **Security Hardening**: Secure path storage, local-only processing - ✅ **Database Management**: Auto-cleanup, persistent storage, state management - ✅ **UX Improvements**: Click-to-enlarge, remove buttons, checkbox persistence - ✅ **Performance**: Optimized image loading with base64 conversion - ✅ **Bug Fixes**: Checkbox state persistence, config loading issues resolved ## 🎯 Integration Workflow & System Architecture ### 🔄 **Complete Integration Flow:** ``` [User Input] → [ai_interaction Tool] → [Tag-based Output] → [AI Agent Rules] → [Enhanced Response] ↑ ↓ └─────────────── [Auto-recall if continue_chat=true] ←─────────────────────────┘ ``` ### 🧠 **Cognitive Enhancement System:** - **Standard Mode**: High-level thinking with 1+ thinking blocks - **Ultra-Enhancement Mode**: 10 breakthrough cognitive modes simultaneously - Quantum Cognitive Mode - Meta-Cognitive Orchestration - Expert Persona Simulation - Time-Dilated Processing - Systems-Level Integration - Psychological Priming Mode - Maximum Cognitive Resource Allocation - Adversarial Self-Testing Mode - Obsessive Quality Standards - Breakthrough Innovation Mode ### 📊 **Output Tag System:** ```xml <AI_INTERACTION_CONTINUE_CHAT>true/false</AI_INTERACTION_CONTINUE_CHAT> <AI_INTERACTION_ATTACHED_FILES> FOLDERS: - workspace_name/relative/path/folder FILES: - workspace_name/relative/path/file.ext </AI_INTERACTION_ATTACHED_FILES> <AI_INTERACTION_WORKSPACE>workspace_name</AI_INTERACTION_WORKSPACE> ``` ## 💡 Advanced Features & Best Practices ### 🎨 **UI/UX Enhancements:** - **Responsive Design**: Adaptive sizing with minimum 800x700 resolution - **Multi-language Support**: Seamless EN/VI switching with persistent config - **Modern PyQt5 Styling**: Semantic color system with button properties - **File Drag-Drop**: Intuitive file attachment with validation - **Context Menu**: Right-click operations for file management - **Debounce Saving**: Smart config persistence with QTimer optimization ### 🔧 **Technical Specifications:** - **Python**: 3.8+ required with PyQt5 dependencies - **Memory**: Minimum 512MB RAM for UI components - **Storage**: ~50MB for tool installation and config - **Platform**: Cross-platform (Windows/macOS/Linux) with native styling - **Performance**: Event-driven architecture with minimal CPU usage ### 📈 **Performance Optimization:** - **Lazy Loading**: Components load only when needed - **Efficient Config**: JSON-based with automatic compression - **Resource Management**: Proper cleanup and memory management - **Caching Strategy**: Workspace state persistence for faster startup ## 🛡️ Security & Privacy ### 🔒 **Security Features:** - **Local Processing**: All file operations are local only, no uploads - **Path Validation**: Robust security checks for file access - **Sandboxed Execution**: Tool runs in controlled environment - **No Data Collection**: Zero telemetry or external data transmission ### 🔐 **Privacy Protection:** - **Config Encryption**: Local config with secure storage options - **File Access Control**: User-controlled file attachment permissions - **Workspace Isolation**: Project boundaries are enforced - **Audit Trail**: Optional logging for security monitoring ## 🌟 System Requirements & Compatibility ### 💻 **Minimum System Requirements:** ``` OS: Windows 10+ / macOS 10.14+ / Ubuntu 18.04+ Python: 3.8 or higher RAM: 512MB available Storage: 100MB free space Display: 1024x768 minimum resolution ``` ### 🎯 **Recommended Setup:** ``` OS: Windows 11 / macOS 12+ / Ubuntu 20.04+ Python: 3.10+ with virtual environment RAM: 2GB available Storage: 500MB free space Display: 1920x1080 or higher GPU: Optional for enhanced UI rendering ``` ### 🔧 **Compatibility Matrix:** | Component | Version | Status | Notes | |-----------|---------|--------|-------| | Python | 3.8-3.11 | ✅ Tested | Recommended 3.10+ | | PyQt5 | 5.15+ | ✅ Required | Core UI framework | | Claude Desktop | Latest | ✅ Optimized | MCP integration | | **Cursor IDE** | **Latest** | **🚀 Recommended** | **AI-first development** | | VS Code | Any | ✅ Compatible | Alternative IDE option | ## 🤝 Contributing **Note**: This is a private repository. Only the owner has push access. For suggestions or issues: 1. Create detailed issue reports 2. Provide reproduction steps 3. Include system information 4. Attach relevant logs or screenshots ## 📚 Documentation & Resources ### 📖 **Documentation Files:** - `rule_for_ai_VI.txt` - Vietnamese agent behavior rules - `rule_for_ai_EN.txt` - English agent behavior rules - `SYSTEM_PROMPT_Claude-4-sonnet-max.txt` - Full system prompt example - `pyproject.toml` - Project configuration and dependencies ### 🔗 **Useful Links:** - [Model Context Protocol (MCP) Documentation](https://modelcontextprotocol.io/) - [Claude Desktop Official Guide](https://claude.ai/desktop) - [Cursor IDE Official Site](https://cursor.sh/) - [PyQt5 Documentation](https://doc.qt.io/qtforpython-5/) - [Python Virtual Environments](https://docs.python.org/3/tutorial/venv.html) ### 💡 **Related Projects:** [- https://github.com/KhaiHuynhVN/mcp-server-agent-comm](https://github.com/KhaiHuynhVN/mcp-server-agent-comm) ## 📄 License & Legal ### 📜 **License:** ``` MIT License Copyright (c) 2025 DemonVN - AI Interaction Tool ``` ### ⚖️ **Legal Notes:** - Tool complies with local processing requirements - No personal data collection - Respects user privacy and data sovereignty - Compatible with enterprise security policies ### 🎯 **Special Thanks:** - Model Context Protocol team for standardized interface - Claude Desktop integration ecosystem - Cursor IDE team for AI-first development tools - Open source Python community - Beta testers and early adopters ### 🔥 **Inspiration:** Project inspired by the need for seamless AI interaction tools with modern UX principles and professional-grade architecture. --- **🚀 Happy Coding with AI Interaction Tool!** *For support, issues, or feature requests, please open an issue on the GitHub repository.*