How many screenshots did you hand your AI today?
Show your agent the screen.
Not your keys.
Screenshots straight into Claude Code, Codex, whatever you use. Passwords get hidden on the way.
Three seconds
Tap ⌘ ⌘
From any app. Or ⌘⇧2.
Drag a region
Space grabs a whole window.
Masked, then pasted
Into the terminal or app you used last.
The details
Masked on your Mac
OCR, plus Apple's on-device LLM where available. No cloud.
Double-tap capture
Faster than a three-key shortcut.
Terminal or desktop app
Claude Code, Codex, their desktop apps, or any app you add.
Paste an earlier shot
⌘⇧1 opens your last 10.
/shots skill
Several shots, one command in Claude Code or Codex.
Text, too
Optionally pastes the OCR text with the image.
Install
Signed and notarized by Apple. macOS 13+.
# Homebrew
brew install --cask woonmayer/tap/yetda
Download Yetda.dmg Open it and drag Yetda to Applications.
Build from source instead
# build git clone https://github.com/woonmayer/yetda.git cd yetda ./scripts/build-app.sh # install (move, don't copy, so macOS permissions stick) rm -rf /Applications/Yetda.app && mv Yetda.app /Applications/ open /Applications/Yetda.app
Asks for Screen Recording and Accessibility on first use.
FAQ
How does secret hiding work?
Yetda reads the capture's text with macOS's on-device OCR, matches known key formats (OpenAI, Anthropic, GitHub, AWS, Stripe, JWT and more), password=… assignments and card numbers, and covers them with a solid mask in the image and in the pasted text. On Macs with Apple Intelligence, Apple's on-device LLM also judges random-looking strings, locally, with no cloud call. If anything goes wrong while masking, Yetda doesn't paste the capture at all. Masked too much? Paste Last Capture Without Hiding pastes the original, which is kept in memory only. Nothing is sent anywhere. It's a safety net, so still glance at what you share.
Why does it need Accessibility permission?
To notice the ⌘⌘ double tap and to press the paste key in your terminal. Without it, Yetda still captures to the clipboard and you paste yourself.
My terminal isn't in the list.
Use that app, then open Paste Into in the menu bar menu and choose Add “<app>”, or pick any app with Add App…. Each app can paste with ⌃V or ⌘V.
Codex (or another app) already uses ⌘⌘.
Turn off “Capture with ⌘⌘” in the menu and use ⌘⇧2 instead.
Captures come out blank or the permission prompt keeps appearing.
Make sure only one Yetda.app exists on your Mac, then run tccutil reset ScreenCapture io.github.woonmayer.yetda, relaunch, and grant again.
Which agents does it work with?
Any terminal agent that accepts a pasted image: Claude Code, Codex, Gemini CLI, opencode and others. Yetda presses ⌃V (⌘V in iTerm2). If your agent wants ⌘V everywhere, like Cursor CLI, set that terminal to ⌘V under Paste Into in the menu. The Claude, Codex and ChatGPT desktop apps are targets too (with ⌘V); any other app can be added from the menu. The /shots skill works in Claude Code and Codex.