Capture text

Capture text is Scratch’s built-in OCR. Click the Capture text button in the footer (or from the export popover), drag a rectangle around any text on your screen, and Scratch drops the recognized text into the buffer at the cursor.

How it works

1. Click Capture text. Scratch dims the screen and turns the cursor into a crosshair.
2. Drag a rectangle around the text you want.
3. Release. The recognized text is inserted at the cursor position.

While capture is in flight, the buffer is locked against new edits so the inserted text lands cleanly. The lock releases the moment recognition completes (typically under a second for a paragraph-sized region).

Permissions

Capture text uses macOS’s screen recording APIs. The first time you trigger it, macOS asks for Screen Recording permission. Grant it in System Settings → Privacy & Security → Screen Recording, then restart Scratch. After that, capture works silently.

What it recognizes

OCR runs locally with Apple’s Vision framework, packaged as a small native binary that ships with Scratch. There is no network call. Recognition handles printed and clean handwritten text in any language Vision supports.

It will not always be perfect on stylized type, low-contrast screens, or dense code. Treat the result as a starting point — re-read it before relying on it.

Privacy

The captured image is never written to disk and never leaves your machine. Only the recognized text reaches the buffer; the bitmap is discarded the moment recognition completes.