Skip to content

Copy the text inside a screenshot on your Mac

When an image enters your library, Clivra runs Apple’s on-device text recognition over it before the image is saved. The recognised words become searchable and can be copied as a text clip. This guide covers the three ways an image can arrive, how to copy the recognised text, and the cases where recognition is incomplete or deliberately skipped.

Three ways an image gets recognised

Copy an image in any app and, with automatic saving allowed and Image enabled under Saved types, Clivra stores it and recognises its text. Import a file with Import files or by dropping it on the Library, and the same recognition runs. Or turn on Save new screenshot files in Settings and choose the folder where macOS writes screenshots; only new files that macOS marks as screenshots are added from that folder.

Capture screen text from the menu bar is different: it asks macOS for a region, recognises the text, and saves a Text clip rather than an image. It needs the Screen Recording permission, which macOS controls.

Find and copy recognised text

Recognised text is part of the clip’s searchable content, so a word visible in a receipt or an error dialog can be found from the Library search or Quick Paste without opening the image. In the image preview, a Recognized text section shows the full result, and Copy recognized text places it on the clipboard as a new Text clip while the original image stays unchanged.

  1. Search for a word you saw in the screenshot; image clips that contain it appear.
  2. Open the clip preview.
  3. Expand Recognized text to read the result.
  4. Click Copy recognized text, then paste where you need it.

Everything stays on your Mac

Recognition uses Apple’s Vision framework with the accurate recognition level and language correction. No image or text is sent to Clivra or any server for this feature. Recognition runs before the image is persisted, so a screenshot that looks like it contains a password, an API key or a card number is rejected by the sensitive-content safeguard and never written to the library when that safeguard is on.

When recognition is wrong or missing

Text recognition is probabilistic. Low-contrast interfaces, small type, handwriting, rotated text and unusual fonts produce gaps or wrong characters, and the language model may autocorrect a code or identifier. Images larger than 100 MB are not processed. If a screenshot contains no text, the Recognized text section does not appear and Copy recognized text is unavailable.

Treat the copied text as a draft to check, not as a certified transcription. Our automated checks recognise a synthetic rendered image; recognition quality on your own varied screenshots has not been separately certified, and results differ between Macs and macOS versions.

Further reading

Apple Developer: Recognizing text in images (Vision)

Prepared with AI assistance and checked against Clivra 1.2.5 source and existing verification records on 6 October 2026. Device-specific results can vary. No fresh hardware certification is implied. Report a correction.