Skip to content

Privacy

What Smarn does with your screen — in plain language.

Last updated October 2026. This describes the current early-access build. If the behaviour changes, this page changes first.

When Smarn looks at your screen

Only when you trigger it — pressing ⌥ Space or holding the push-to-talk key. There is no always-on recording and no background capture. The capture is a single window: the one that was in front when you pressed the key, never Smarn's own windows.

What is captured

What is scrubbed before anything is sent

API keys, access tokens, JWTs, private keys and password=…-style credentials are replaced with a [REDACTED] marker inside the app, before the request is built. Redaction is code, not a prompt.

Where it goes

To the AI provider you connect, using your API key, under that provider's terms — currently Anthropic. Your key is stored in the macOS Keychain and read only to make the request; it never reaches the interface or any log. Smarn has no server of its own in this build: nothing is sent to DataOrc.

What stays on your Mac

Memory. Every capture you ask about is kept in ~/Library/Application Support/Smarn as an image file plus a database row with the redacted recognised text, so you can search what you've seen. Memory is never uploaded. You can turn it off, forget one capture, or delete everything from Settings.

What Smarn never does

Permissions it asks for

Screen Recording and Accessibility are required for the core loop. Microphone and Input Monitoring are only for voice and are optional. The onboarding window explains each one and lets you grant or skip it.

Questions

Smarn is made by DataOrc LLP. Release notes and downloads: github.com/dataorchestration/smarn-releases.