Email: security@midtowntg.com

This tool processes local screenshot files:

• Input: PNG/JPG files from ~/Pictures/Screenshots/
• Processing: OCR via Azure AI Vision API
• Output: Markdown SOPs in ~/Documents/SOPs/

• API calls use HTTPS/TLS 1.2+
• Images are transmitted to Azure for OCR processing
• Microsoft does not store or train on your images (Azure AI Vision privacy)
• API key stored in environment variable or config file

Config file at ~/.config/sop-generator/config.yaml:

azure:
  api_key: "${AZURE_VISION_KEY}"  # Use env var, don't hardcode
  endpoint: "https://your-resource.cognitiveservices.azure.com"

Never commit API keys to git.

• Uses Windows clip.exe command via subprocess
• Input text is sanitized before clipboard copy
• No shell=True (secure subprocess usage)

Generated SOPs may contain:

• System configurations
• Internal network details
• Sensitive UI elements

Review before sharing externally.

Recommended Azure permissions:

• Cognitive Services User role only
• Single resource: Computer Vision
• No access to other Azure resources

• Screenshot paths validated as existing files
• No shell expansion of user input
• Image format validation (PNG, JPG only)
• File size limits enforced

When contributing:

1. Never commit API keys
2. Use environment variables for secrets
3. Test with dummy images, not production screenshots
4. Validate all file paths