Overview

Cursor Crane

Most hint-based Mac tools stop at UI elements. Cursor Crane extends the same interaction model to windows, multi-window workflows, and text editing, and it uses predictable hints rather than arbitrary ones.

The interaction is always the same three beats:

  1. Look at your target.
  2. Choose the action you want (move the cursor, click, drag, scroll).
  3. Type the hint that performs it.

Start here#

If you want to Read
Install the app and grant permissions Setting Up Cursor Crane
Decide how you will drive it day to day Choosing a Usage Style
Edit settings and TOML config Configuration
Get back to normal typing Exit and Temporary Exit
Understand the modes Modes overview
Automate a repeated task Using Scripts
Solve a specific problem Use Cases

The claw machine#

When you press your activation shortcut, Cursor Crane starts what the app calls the claw machine. It opens in Command Mode, where letter sequences trigger actions and single commands take you into the other modes.

While the claw machine is active, most system and app shortcuts keep working. Command + C still copies, Command + W still closes a window, and arrow keys still reach the foreground app.

The eleven modes#

Each mode solves one class of pointing problem:

  • Command Mode enters command sequences and reaches every other mode.
  • Mouse Control Mode simulates movement, clicks, scrolling, and gestures.
  • Element Mode jumps between clickable elements by hint.
  • Grid Mode reaches any pixel through progressively smaller grids.
  • Window Mode switches windows and menu bar items.
  • Find Mode finds visible text with OCR and moves the cursor to it.
  • Input Mode types normally and adds text-editing shortcuts.
  • Scroll Mode picks a scrollable region and scrolls it.
  • Adjust Mode moves and resizes windows, splitters, and sliders.
  • Element Menu Mode inspects a target, copies from it, and captures screenshots.
  • Element Burst Mode pulls text and structured values out of one element.

Scripting#

When a workflow is not built in, write it. A script is a .js file with a meta object and a run(ctx) function; it can inspect accessibility elements, run shell commands, and present native forms. See Using Scripts, Element Selectors, and the Script API.

Plans#

Core navigation is free and needs no license. Scroll Mode, Find Mode, Adjust Mode, Element Burst Mode, app-specific configuration, and the richer Input Mode shortcuts are Unlocked features. A 14-day full-feature trial can be enabled inside the app. See Plans and Licensing.

Reference#

Community#

Questions and bug reports go to the Discord community or support@expressionloss.com.

Updated