🐛 Bug Report

Describe the bug

The feature to detect the screen context using OCR doesn't work. Output looks like this.

<CURRENT_WINDOW_CONTEXT>
Active Window: StatusIndicator
Application: Window Server
Window Content:
No text detected via OCR
</CURRENT_WINDOW_CONTEXT>

Tried this in multiple apps. Even tried to reproduce your Gmail example.
I'm using an M1 Macbook Pro.

Expected behavior

Current window context sent to model.

Additional Context

• Frequency: Happens every time.