OpenSSDocumentation
DesktopLocalPro

Desktop capture and OCR

Capture supported macOS windows and submit extracted context for analysis.

Updated 22 July 2026Applies to 0.2.8
On this pageSupported targetsCaptureImage-first analysisCapture from mobileCapture privacy

Supported targets

TargetBehaviour
ChromeCaptures the target Chrome window.
PowerPointCaptures the window or requires Slide Show mode for full-slide capture.
WordCaptures the supported Word window; CLI also exposes document automation.

Capture

  1. 1
    Prepare the target

    Open the supported app and keep it on the display where OpenSS expects to capture.

  2. 2
    Choose Capture

    Select the target and full-slide behaviour where shown.

  3. 3
    Allow permission

    Approve Screen Recording when macOS asks, then retry if the first permission change requires it.

  4. 4
    Review

    OpenSS extracts text locally and sends the submitted prompt and relevant image/context through the selected AI route.

Image-first analysis

Screenshot requests always include the original PNG or JPEG pixels as well as OCR. The vision-capable model receives the image as primary evidence, with the user instruction and locally extracted OCR clearly separated as supporting text. Diagrams, charts, colours, selected states and spatial layouts are therefore available to the model even when OCR is empty or incomplete.

Capture from mobile

Capture privacy