Context capture
dooze gathers context from what you are working on, so you do not have to explain everything from scratch.
Three capture systems
Text selection — Select text in any app. dooze reads it through the macOS Accessibility API. No copy required.
Clipboard — When you press Cmd+C, dooze captures the copied text or image. You see a brief coral-to-purple pulse on the panel as confirmation.
Screen capture — In higher capture modes, dooze can read text from your screen using OCR or send a screenshot to the AI model for visual understanding.
Context pills
Captured items appear as pills below the input field. Each pill expires after 10 minutes. Click the X on any pill to remove it early.
Four capture modes
These are additive. Each mode includes everything below it.
| Mode | What it does | Requires |
|---|---|---|
| Off | No automatic capture. You can still paste or drag content in | Nothing |
| Cursor | Captures text you select in any app. This is the default | Accessibility |
| Cursor + Screen Text | Adds one-shot OCR — dooze reads visible text from your screen | Screen Recording |
| Cursor + Screen Text + AI | Adds screenshot-to-AI — dooze sends a screenshot for visual understanding | Screen Recording |
The higher modes (Screen Text, AI) capture once per activation, then automatically drop back to Cursor. They are designed for "grab this screen" moments, not continuous monitoring.
To change your capture mode, click the capture icon in the bottom bar of the panel.
Privacy exclusions
Password managers (1Password, Bitwarden, LastPass, Dashlane, KeePassXC, Enpass, RoboForm, Keychain Access) are excluded from all capture. Browsers are excluded from clipboard capture only. Secure input fields like password boxes are never captured.
See Privacy for the full list.