At a glance
- One box, every content type: links, text, code, rich text, files and images, across the whole local history.
- Words inside screenshots: on-device recognition indexes the text in each copied image.
- Keyboard only: open the popup, type, arrow to the clip, release or press Return to paste.
- Filters in the History window: content type, date range, and the app you copied from.
- Nothing leaves the Mac: search and text recognition both run locally, offline.
Finding the words inside a screenshot
Recognition is retroactive by design. You did not have to think about grabbing the text when you took the screenshot, which is the whole point, because nobody does.
Once you have found the image, ⇧Enter pastes the recognised words instead of the bitmap. Plain ⌘V or Enter still pastes the picture, and the image stays in history either way.
It happens at copy time, off the copy path. Klipto hands the pixels to Apple's Vision text recognition in the background and stores the result next to the image. Pressing ⌘C never waits for it. Weeks later, three words you remember from an error dialog find the screenshot of that dialog.
It reads what you copied, and nothing more. Klipto does not watch your screen and does not capture windows you never touched. If an image never entered your clipboard, it is not in the history and not in search.
Klipto is not alone here. Raycast's clipboard history and ClipBear both do text recognition on images too, so image search is a reason to run a clipboard manager rather than a reason to pick mine. What I would compare instead: how long the history lives, what you can filter it by, and what the app can do to a clip on the way out.
Two surfaces, one history
The popup is the fast lane. Hold or tap ⌥⌘V, type, and it filters a small in-memory buffer of recent items plus your snippets. Snippets match on their body text, not only their name, so a saved block you titled badly is still findable.
The History window covers the archive. It pages straight from the local database rather than building an index in memory first. That is why it opens fast after tens of thousands of copies. Search there sits next to filters for content type, date range and source app. Image bytes live in a de-duplicated store outside the database, so ten copies of one screenshot cost one file.
Once you have the clip, the preview panel can reshape it before it lands, and the Stack can collect several and paste them in order.