A Windows-only Model Context Protocol (MCP) server that enables AI agents to capture screenshots of applications or the entire screen, with optional visual question answering through local or remote AI models.

• Full Screen Capture: Capture screenshots of the entire screen or specific monitors
• Window-Specific Capture: Target specific application windows by title or process name
• Window Enumeration: List all visible windows with their process information
• AI-Powered Analysis: Analyze screenshots using OpenAI, Anthropic Claude, or local models
• Multiple Image Formats: Support for PNG and JPEG output
• Windows Integration: Deep Windows API integration for reliable window targeting

• Windows operating system (Windows 10/11 recommended)
• Python 3.8 or higher
• Required Python packages (see requirements.txt)

1. Clone or download this repository

2. Install dependencies:

   pip install -r requirements.txt

3. Set up API keys (optional, for AI analysis):

   # For OpenAI
   set OPENAI_API_KEY=your_openai_api_key
   
   # For Anthropic Claude
   set ANTHROPIC_API_KEY=your_anthropic_api_key
   
   # For local models (e.g., Ollama)
   set LOCAL_MODEL_URL=http://localhost:11434

Run the MCP server:

python screenshot_server.py

Capture a screenshot of the entire screen or a specific monitor.

Parameters:

• monitor (integer, optional): Monitor number (0 for primary, 1+ for additional)
• format (string, optional): Image format ("png" or "jpeg", default: "png")

Capture a screenshot of a specific application window.

Parameters:

• window_title (string): Title or partial title of the window
• process_name (string): Process name (e.g., "notepad.exe")
• format (string, optional): Image format ("png" or "jpeg", default: "png")

Note: Either window_title or process_name is required.

List all visible windows with their titles and process names.

Parameters: None

Analyze a screenshot using AI and answer questions about it.

Parameters:

• image_data (string): Base64 encoded image data
• question (string): Question to ask about the image
• model_provider (string, optional): "openai", "anthropic", or "local" (default: "openai")
• model_name (string, optional): Specific model name (default: "gpt-4-vision-preview")

• OPENAI_API_KEY: Your OpenAI API key for GPT-4 Vision analysis
• ANTHROPIC_API_KEY: Your Anthropic API key for Claude analysis
• LOCAL_MODEL_URL: URL for local model API (default: http://localhost:11434)

• gpt-4-vision-preview
• gpt-4o
• gpt-4o-mini

• claude-3-sonnet-20240229
• claude-3-haiku-20240307
• claude-3-opus-20240229

• Any Ollama model with vision capabilities (e.g., llava, bakllava)
• Custom local vision models with compatible API

# Capture entire screen
{"tool": "capture_screen", "arguments": {}}

# Capture specific monitor
{"tool": "capture_screen", "arguments": {"monitor": 1, "format": "jpeg"}}

# Capture by window title
{"tool": "capture_window", "arguments": {"window_title": "Notepad"}}

# Capture by process name
{"tool": "capture_window", "arguments": {"process_name": "chrome.exe"}}

# Analyze with OpenAI
{
  "tool": "analyze_screenshot",
  "arguments": {
    "image_data": "base64_encoded_image_data",
    "question": "What applications are visible in this screenshot?",
    "model_provider": "openai"
  }
}

# Analyze with local model
{
  "tool": "analyze_screenshot",
  "arguments": {
    "image_data": "base64_encoded_image_data",
    "question": "Describe what you see in this image",
    "model_provider": "local",
    "model_name": "llava"
  }
}

• This server requires Windows API access and can capture sensitive information
• Screenshots may contain private data - ensure proper handling
• API keys should be stored securely and not committed to version control
• Consider network security when using remote AI models

1. Import Error for Windows modules: Ensure you're running on Windows
2. Permission denied: Run as administrator if capturing system windows
3. Window not found: Check window titles with list_windows tool first
4. AI analysis fails: Verify API keys are set correctly

Enable debug logging by modifying the logging level in the script:

logging.basicConfig(level=logging.DEBUG)

MIT License - see LICENSE file for details.

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Test on Windows
5. Submit a pull request