Home > Mac

How macOS App Sandboxing Protects Users

By Alexander Fox
How macOS App Sandboxing Protects Users Featured Image

Introduced in 2007 and required by 2012, sandboxing is a tool used by macOS to limit the damage that a hijacked app can do. Apple says, “While App Sandbox doesn’t prevent attacks against your app, it does minimize the harm a successful one can cause. macOS app sandboxing protects users by limiting how much trouble an application can cause.

Also read: 4 of the Best Sandbox Applications for Windows to Keep You Safe

What is macOS App Sandboxing?

Each app gets its own area to play in: a “sandbox.” If the application wants to reach outside its sandbox, it needs to ask the operating system for permission. Depending on the sandbox settings, the OS will either deny or approve the application’s request while providing the least specific information possible to complete the request.

about-sandboxing

Think of a car. The driver is protected by multiple security features like anti-lock breaks, air bags, crumple zones, and more. In 2007, the time of App Sandboxing’s introduction, no analogous computer systems existed. As the introductory presentation asked, where were the seat belts for computers? Sandboxing protects users like cars protect their drivers: meeting failures with damage-reduction systems. Whether caused by malicious activity or coding errors, damage goes down.

Because sandboxing limits what apps can do, it can restrict developer freedom. Sandbox apps run more slowly and take longer to develop. Thanks to the Mac’s enormous capability, sandbox limits can have a huge impact on the compatibility of various apps. As a result, power users are often driven to choose apps that run outside the sandbox, either for performance or feature issues.

How Does App Sandboxing Work?

Sandboxing is based on the principle of least privilege. In short, systems can do what they need to do but no more. By limiting each part of a system to completing only its declared goal, you reduce the chance of your app being hijacked. For example, there’s no reason for a flashlight app to have access to your contacts list.

Apps can reach outside their sandbox but only with operating system permission. Take the “Save and Open” dialog box in macOS. The app, inside its sandbox, cannot directly access filesystem resources on your hard drive. It cannot, for example, draw an open panel at “~/Documents.” Instead, the app must ask the Powerbox API with NSOpenPanel and NSSavePanel classes to access the panel.

macos-app-sandboxing-powerbox-design

The application cannot see what’s happening inside Powerbox directly. Only the opened or saved file will be accessible to the app. This way the apps can perform critical functions without unnecessary risk.

That functionality is enabled by an entitlement (specifically com.apple.security.files.user-selected.read-write). App developers set entitlements which declare what an app does. Based on the declared entitlements, the operating system permits the application an appropriately-limited level of functionality.

This thought process underlies the entirety of the app sandboxing model and mechanism: apps must declare intention and ask permission from an OS-level boss to accomplish anything dangerous.

Sandboxed Apps vs. Non-Sandboxed Apps

Since June 1st, 2012, all third-party applications distributed through the Mac App Store must be sandboxed. While sandboxing does permit a large range of app functionality, you’ll find that Mac App Store apps are often more limited than their non-sandboxed components. Some developers even maintain two versions: a fully-featured app for direct download and a gimped version for the Mac App Store. Thanks to the greater complexity of sandbox development, adding new functionality to an application is more difficult – if that functionality is even permitted by the OS, that is.

While sandboxed apps can get the benefit of Mac App Store distribution, you can guess how valuable we think that is. Sandboxing can also be extended with security permissions. While an app cannot turn on Accessibility permissions for itself, it can ask the user to do so.

macos-security-privacy-permissions-hero

To see which of your apps are sandboxed, open Activity Monitor, and right-click the column titles to add “Sandbox” to the window.

macos-app-sandboxing-activity-monitor-view

There are some apps that can simply never exist in a sandbox. In fact, a variety of valuable use cases are prevented by sandboxing. Sandboxing prevents inter-application communication, observation, or modification, significantly limiting how applications can interact. System-wide shortcuts like TextExpander are totally prohibited, since that level of functionality could be permitted by the sandbox.

Conclusion

While promising, macOS app sandboxing wasn’t executed well. It limited the unique selling points of Mac apps, like speed and enhanced functionality. Apps running outside the sandbox are almost always more capable and faster. From this power user’s perspective, my most-used apps are non-sandboxed. Apps like TextExpander, SnagIt, and TotalFinder are all crucial to my daily work. To avoid similar widespread avoidance, future security systems need flexibility and power balanced with transparency.

Image credit: Apple

Subscribe to our newsletter!

Our latest tutorials delivered straight to your inbox

Alexander Fox Avatar
Alexander Fox
Alexander Fox is a tech and science writer based in Philadelphia, PA with one cat, three Macs and more USB cables than he could ever use.

Read next

A Greenland shark caught in 2016 was carbon-dated to roughly 400 years old, meaning it was already swimming the Arctic when Shakespeare was writing Hamlet — and biologists still can't explain how its cells resist ageing
In 2026, scientists assembled the Greenland shark’s 5.9-billion-letter genome and examined retinas from animals more than a century old, finding intact vision, expanded DNA-repair pathways and unusual chromatin machinery inside a vertebrate believed capable of surviving for nearly four centuries
At 3 gigahertz, a processor has only a third of a nanosecond between clock ticks, enough time for light to travel about 10 centimetres in vacuum before real interconnects, logic gates and capacitance shrink the distance a signal can cross on the chip
When Adobe announced Flash’s end-of-life in July 2017, it closed a technology that had once run on 99 per cent of desktops — and took with it an estimated four million animations, games and interactive works now preserved only by an amateur archive called Flashpoint
When Japan’s trade ministry put the PlayStation 2 under export control in April 2000, its 6.2-gigaflop hardware was considered adaptable for missile guidance, briefly forcing anyone exporting more than one console to seek government approval
Every QR code hides three larger squares in its corners — they let a scanner find the code at any angle, and the pattern also carries enough redundancy to still read after roughly 30 per cent of the code is damaged or covered
Standard QR codes hide three larger squares in their corners so scanners can orient them from any angle, while their strongest Reed-Solomon setting can restore roughly 30 per cent of damaged codewords, though not necessarily 30 per cent of the visible square
In 1911, the Dutch physicist Heike Kamerlingh Onnes cooled a thread of mercury to 4.2 kelvin in a bath of liquid helium and its electrical resistance vanished completely, and a current set circulating in a superconducting loop like that will still be flowing years later, indistinguishable from the day it began
Red, green and blue became the primary colours on every screen because the human retina carries just three types of cone cells — the pixels aren't matching the light, they're matching your eyes
Red, green and blue became the primary colours on every screen because the human retina carries just three types of cone cells — the pixels aren’t matching the light, they’re matching your eyes
Inside a modern processor, transistors switch on and off billions of times a second — and they've shrunk to about 3 nanometres wide, roughly 15 silicon atoms across, close to the physical limit of how small a switch can be
Inside a modern processor, transistors switch on and off billions of times a second — and they’ve shrunk to about 3 nanometres wide, roughly 15 silicon atoms across, close to the physical limit of how small a switch can be