Documenting a workflow means screenshots, cropping, arrows and typing out every step. StepGrab does it for you: hit record, do the task once, and every click comes back as an annotated screenshot with the step already written. Fix anything in the editor and black out private details before you share. Mac and Windows, on-device, no account, nothing uploaded. Export as PDF, clickable HTML, Markdown, GIF or MP4. Free for up to 15 steps. Pro $2.99/mo, $12.99/yr or $44.99 once.
This is the 2nd launch from StepGrab. View more
StepGrab
Launching today
New since our first launch: StepGrab runs on Windows, and projects open on Mac and PC. Black out emails, names or prices, and the box is burned into every export so nobody can lift it off. Every guide opens in an editor: rewrite steps, add arrows, boxes and text, reorder, undo. The direct Mac download reads the real name of every button you click. Still the core: every click becomes a screenshot, arrow and sentence. No account, nothing uploaded. 5 free Lifetime licenses: see the comments :)
Free Options
Launch Team / Built With
AWS Startup AdvisorPersonalized AWS guidance that meets you where you’re already building
Promoted
Thank you, and your second paragraph is the reason the app has the shape it does. A lot of what I document has account settings in it too, which is why I never built a server. There's nothing to upload because there's nowhere for it to go.
On the unlabeled icons, you've found the real weak spot. The step text comes from reading what's visible on screen plus whatever label the control exposes. When a toolbar icon has neither, there's nothing left to describe and it falls back to something generic.
The clean fix would be the Accessibility API, because an unlabeled icon usually still carries a description the developer set. That's where it gets murky. Apple restricts that API on the App Store to genuine accessibility use, so I work with a narrower permission instead. I could ship outside the App Store and have full access tomorrow, but then the privacy claim becomes "trust me" rather than something the sandbox actually enforces. For your kind of content that's the wrong trade, so I'm not making it.
The route I'm taking instead is the on-device vision model coming in macOS 27, which can look at the icon itself rather than asking the app what it's called. That's the 2.0 plan. Slower, but it doesn't cost you the thing you came for in the first place.
Thanks for writing this up properly. It's more useful than most of what I get.