👁️ visual-assertion-engine (v1.0)

A "vision-enabled" assertion engine that provides real-time feedback loop for OS automation. It reads the latest screen frames from the rolling buffer and uses OCR to detect state changes.

🚀 Capabilities

1. Text Detection: Searches for specific keywords on the screen (English/Japanese).
2. State Waiting: Loops until a condition is met or a timeout occurs.
3. Low Latency: Uses the existing rolling buffer (10 frames) maintained by the daemon.

📦 Usage Examples

# Wait until "Build Success" appears on the screen
gemini run visual-assertion-engine --text "Build Success" --timeout 60000