Captura DeskGet Captura Desk

GETTING STARTED

Your first useful document.

From a local recording to a guide you can share. Start with a short task and a few non-sensitive screens.

01 / SET UP

Choose how to install.

Windows installer

Download version 0.2.0 for Windows x64 from the download page. Run the installer for your Windows account, then open Captura Desk from the Start menu. Git and Node.js are not required.

This beta is unsigned, so Windows may show an unknown-publisher or SmartScreen warning. Review the linked release notes before deciding whether to install it.

For an update, close Captura Desk and run the newer installer. Automatic updates are not available yet. Back up your workspace before upgrading.

Run from source

Install Git and Node.js 22.18 or newer on Windows, then run:

Windows terminal
git clone https://github.com/capturadesk/captura-desk.git
cd captura-desk
npm ci
npm start

The app opens after the build finishes. For development with live reload, use npm run dev.

Mixed-DPI displays, elevated applications, and long sessions still need broader testing.

02 / CAPTURE

Record a task you know.

  1. Create a project in your workspace. Add instructions describing the document you want.
  2. Choose New recording, name the task, and select the display. Use Identify displays if you are unsure.
  3. Start recording and work in the selected display. Clicks save screenshots. Use Capture now or Ctrl + Shift + S for a manual capture during recording.
  4. Pause before showing anything sensitive. Choose Finish when the task is complete.

Review the captured steps and both before/after images. The before frame is sampled; it is not guaranteed to show the exact instant before your click.

03 / ADD DIRECTION

Generate a draft, if you want one.

  1. Open AI providers. Save your OpenAI or Anthropic API key and test the connection.
  2. In Workspace settings, select a provider and a model supporting images and structured output.
  3. Choose Generate documentation. Review the captures, exclude sensitive ones, and confirm which provider receives them.
  4. Review the draft and choose Save as new document to preserve the original.

Up to 200 selected captures can be processed in batches. Longer recordings make multiple requests and may take longer or cost more. Provider API billing is separate.

To redact an image, enlarge it and choose Annotate screenshot. Save edits for both frames before AI generation or export.

04 / REFINE

Make the draft your own.

Click a description to edit its Markdown. Click away to see headings, lists, and tables formatted. Changes save automatically.

Ask AI for a change such as "Make this shorter". Text refinement sends the document text and your instructions, without uploading screenshots again. Always check the result against the evidence.

Use the document selector to move between the original and revisions. Screenshot edits apply across revisions that share the same capture.

05 / HAND IT OVER

Share one file.

Select the revision you want, then choose Export > HTML - single file. The file includes formatted text and screenshots with saved annotations. It opens offline in a browser.

For Markdown, choose Markdown with images and share the image folder with the document. Exports use the selected revision.

Keep a separate workspace backup through Workspace settings > Back up workspace. A shareable document is not a backup of the project.

Complete app documentation