Your screen is full of unselectable text. Click it, select it, copy it — mouse-free.
Screenshots, videos, error dialogs, someone else's screen share, a PDF that's secretly an image — the words are right there, but the mouse can't touch them. Textracto reads your whole screen; type a few letters to find any line, then one keystroke clicks it, selects it, or copies it — right in the app it lives in, entirely from the keyboard.
Twenty seconds, one real screen.
An error code lifted straight out of a macOS alert — text the system gives you no way to select — then the dialog's button clicked, without the mouse. Nothing here is a mock-up.
Rather not load YouTube at all? Play the file straight from this site — no third party involved.
If you can see the text, you can act on it.
It's screen OCR for macOS with one difference: where other tools stop at copying text from a screenshot, Textracto keeps going — it finds the words anywhere on screen, then clicks, selects, or replaces them right in the app they live in. So yes — if you're looking for a TextSniper alternative, or just a way to copy text from a screenshot on a Mac without the mouse, it does that on the way past.
Click anything by name
Find a button, link, or cell by the words on it and press ↩ — Textracto brings that window forward and clicks exactly there. ↩↩ double-clicks, ⌃↩ right-clicks, and holding ⌘/⌥/⇧ passes the modifier through.
No text? No problem.
The thing you want has no label — a checkbox, a toolbar icon? Find the text beside it, then press ⇥ to hop the target onto the nearest control and click that. Each ⇥ steps outward.
Select, replace, cut
Mark a span with .* and ↩ selects it right in the app. ⌘V replaces it with your clipboard; ⌘X cuts it — editing text you were never able to select.
Find & copy, fzf-fast
Type a few letters; fuzzy matching ranks every line live, matched characters highlighted. ⌘C copies the match to your clipboard — the classic, mouse-free.
Grab exactly the right piece
Matches stop at punctuation, so legal picks out legalEntityId and nothing else. Double-tap ⌘ to start selecting it on screen, then tap again to widen — the enclosing quotes, then the braces, then the whole line, one balanced pair at a time.
One window or all of them
⌥Space reads the window you're looking at. ⌥⇧Space reads every window on screen into one searchable list, each line tagged with the app it came from.
Or just drag a box
⌥⌘S dims the screen to a crosshair — drag a rectangle and every line inside it lands on your clipboard as one clean line, newlines stripped. No palette, no typing.
Reads 30 languages
Japanese, Chinese, Korean, Russian, Arabic and two dozen more. Name your language in Settings or let Textracto work out the script itself — and keep English alongside it, for the menus, code, and file paths that stay English anyway.
From “I can see it” to done, in three keystrokes.
Summon ⌥Space
The screen is captured and OCR'd on-device in a blink. The palette appears with every visible line of text, tagged by window.
Type to find it
Fuzzy matching ranks the candidates as you type — matched characters highlighted, best match on top, just like fzf.
Choose what happens
↩ clicks it; ↩↩ double-clicks, ⌃↩ right-clicks. ⌘C copies the text, ⌘X/⌘V cut and replace it. ⌘⌘ switches ↩ to selecting, and widens as you keep tapping. Add .* and ↩ selects from the start — and you never touched the mouse.
$2.49 during the beta. Lock it in before it's $4.99.
- Every feature, nothing locked
- 250 actions on the house — copies, clicks, and edits all count
- After that, a small red reminder appears — the app keeps working forever
- Silences the reminder
- All future updates included
- Offline license key — verified on your Mac, never phones home
- The warm glow of supporting a tiny tool early
Textracto needs two macOS permissions and nothing more: Screen Recording, to read the pixels, and Accessibility, to read exact text from apps that publish it and to move the pointer and type on your behalf. Both stay on your machine — capture and recognition are entirely on-device, images are discarded the moment OCR finishes, and the app has no account, no telemetry, and no analytics. The only request it ever makes is a once-a-day update check that sends nothing about you — and you can switch that off too. Read the privacy policy →