"Searchable screenshots" is one of those features that sounds simple until you've lived without it. You take a screenshot of an error message, a receipt, a Slack thread, a slide from a deck — and two weeks later you remember exactly what it said but have no idea what you named the file, or whether you named it at all. ScreenClipr's answer is to never make you name it: every capture is read and indexed automatically, and you search by what's actually in it.
What happens the moment you save a capture
As soon as a screenshot or an imported image lands in your library, it's handed to Apple's Vision framework for on-device text recognition — the same OCR engine that powers Live Text elsewhere in macOS, running locally on your Mac with no network call involved. The recognized text is written into a local search index alongside the capture's filename, any tags you've applied, and the source app it was captured from (for window captures, ScreenClipr already knows which app owned the window, so that's indexed too).
This isn't limited to captures you take inside the app. Drag an existing PNG or MOV into the library to import it, and it gets copied in and OCR-indexed exactly the same way. And if the app finds capture files on disk from before they were indexed, a background pass catches them up automatically the next time it launches — so your existing history doesn't get left behind.
The index itself
Under the hood, the recognized text is stored in a local full-text search index (SQLite's FTS5), which is what makes search fast rather than a slow per-file scan — results update as you type, even across a large library. When you type into the library's search field, results aren't ranked as one flat list: a match on the filename or an assigned tag ranks above a match that was found only inside the OCR'd text, so searching "invoice" surfaces a screenshot you explicitly tagged "invoice" before it surfaces an unrelated capture that merely happens to contain the word somewhere in its text. Under each result while you're searching, a short snippet shows you exactly where the match was found, so you're not guessing which of five similar thumbnails is the right one.
Search without saving anything at all
Sometimes you don't want a new library entry — you just want the words. Grab Text (OCR), available as a menu-bar action, a global shortcut (Shift+Cmd+T), and a one-click button in the editor toolbar, runs the same on-device recognition and puts the plain text straight on your clipboard — useful for pulling a phone number off a photo, or copying a paragraph out of a PDF that doesn't let you select text normally.
Why this stays entirely on your Mac
Because the recognition runs locally, nothing about a capture — or a search — ever has to leave your Mac to work. There's no cloud OCR service being called, no account tying your search history to a server somewhere, and no dependency on being online for search to work at all. That's a deliberate design choice, not an incidental one: it's the same reason the rest of ScreenClipr's pipeline (capture, annotation, storage) is local-only. Your screenshots are frequently the most sensitive things on your machine — error dialogs with real data in them, financial documents, internal chats — and a searchable archive of that shouldn't require trusting a third party with it.
What this looks like in practice
- You screenshot a terminal error at 11pm, forget about it, and six weeks later search the exact error string — it's there, with a snippet showing the match.
- You screenshot a dozen receipts over a trip and later search a vendor name to pull just those up, without having sorted or renamed a single one.
- You're in a meeting, someone shares a phone number or a code in a screen-share, you screenshot it and Grab Text it directly onto your clipboard without ever opening the editor.
None of this requires you to have planned ahead — you don't tag or file anything at capture time. The indexing happens automatically, in the background, and the search just works when you eventually need it.