Copy text and code from screenshots

Neither way changes the screenshot. Reading costs 1 credit per image; laying out code or fixing spelling costs 3 more.

Before

The first lines of cart.py in a code editor, with line numbers.

After

The same lines laid out as code, as the workflow's formatted result shows them in Code view.

What you get

• A screenshot's words as text you can paste.
• Code laid out with its indentation, without line numbers, or text with its typos fixed.

Pick the tool

• Code laid out, or text with its typos fixed: the Agent, in a chat.
• At scale: a workflow, which treats every screenshot of a run the same way.

Code or fixed spelling: the Agent

Attach the screenshot in a new Agent chat and send:

Code shows the laid-out version, Original the text as read; Copy the text copies the one you see. For labels or notes, send this instead:

Formatting uses 3 credits, plus 1 to read an image not read yet. See Copy code from a screenshot.

At scale: a workflow

• Code or fixed spelling: Images → Format → Save, with Format as on Code or Spelling fixed; 4 credits per screenshot. See Turn code screenshots into formatted code.
• Text as written: Images → OCR → Save. Start from scratch (Sort photos into folders) and choose AI → OCR for the empty step; 1 credit per screenshot.

Text results don't go to the Library: copy a result's text, Download a .txt file, or save each to Google Drive.

What to check

Compare names, numbers, prices and, in code, indentation and brackets with the screenshot; blurred or angled writing reads less reliably. A formatted result's list of changes, such as 12 changes, shows what the formatting changed.

Common mistakes

• Naming a language. Ask for “code”; “Format it as Python code” reads as a request to convert, which the Agent turns down.
• Putting OCR before Format. Format reads the screenshot itself, so OCR → Format pays for the read twice.

Open copyimage.ai and try it on a screenshot of your own, or on the Code in VS Code example in a new Agent chat.