Bigarrow is a new macOS command-line tool that lets AI agents visually point to screen elements with arrows, boxes, and text. Released on GitHub, the lightweight Swift binary creates click-through arrows that highlight buttons or dialogs without stealing focus or requiring permissions. It addresses a common limitation where AI agents can identify UI elements but cannot interact with them directly, such as clicking “Allow” on permission prompts.
Developed by Franz Enzenhofer, bigarrow works by overlaying a large, friendly arrow on top of everything on the screen, pointing at the specified element and then disappearing. The tool requires no daemon, menu-bar icon, account, or telemetry, and it functions across all displays and Spaces on macOS. It is designed to assist AI agents that perform tasks but need a human to complete certain actions like permission approvals or two-factor authentication confirmations.
This tool is significant because software agents increasingly perform complex tasks on Macs but face roadblocks when human interaction is required for security or privacy reasons. Bigarrow bridges this gap by visually guiding users to the exact UI element needing attention, improving the workflow between AI automation and human input. It complements AI capabilities without automating sensitive clicks, respecting macOS security constraints.
Bigarrow is available under the MIT license on GitHub, enabling developers to integrate it into their AI workflows. The project page at github.com/franzenzenhofer/big-arrow-on-the-screen provides the source code and usage instructions for this minimal macOS CLI utility.