# I refused to give a launcher Full Disk Access, so I built one > Source: > Published: 2026-08-17 13:54:44+00:00 August 17, 2026 # I refused to give a launcher Full Disk Access, so I built one Commercial software in closed ecosystems is running out of road. The App Store is full of launchers that want to be operating systems, one of them told me to uninstall it if I would not hand over Full Disk Access, so I built the launcher I actually wanted in half an hour and put it on GitHub under MIT. TL;DR: I refused to give a launcher Full Disk Access, so [I built one in half an hour]. Commercial applications on Apple’s App Store, and in every other closed ecosystem, are going to die sooner than their authors think. Most of them are already obsolete, stuffed with features that have nothing to do with how anybody works in 2026. And the best part is that if you want an app for macOS and you don’t feel like paying for it, you can call daddy ChatGPT Sol or uncle Claude Code and have the app of your dreams in a matter of minutes. That is the future, and it arrived while the App Store was busy filling up with garbage. Because that is what it is. Apple managed to infest its own store the moment it let developers charge for their software, and what was sold to us as a walled garden now looks a lot more like a digital wasteland. Hundreds of thousands of applications, and maybe the top one percent are good enough to be worth money. Don’t misread me here: I am not against paying for software. I have paid for every application that genuinely earned a place in my life, and I have donated to developers who gave their work away for free, because the effort and the care that goes into good software is real and I value it. But a garden with a wall around it and no gardener eventually becomes a cesspool, and that is roughly where we are: a place where hundreds of thousands of mediocre developers flood the ecosystem with low quality work and drown out the people doing it properly. The practical consequence is that finding an application that does one thing well has become genuinely hard. Everything wants to do everything. Imagine downloading `ls` , a command that lists the contents of a directory, and then discovering that `ls` also takes notes, sends email, and syncs your calendar. You would assume something had gone badly wrong in that project’s history, and you would be right. He who chases two rabbits catches neither, and the Unix philosophy has always struck me as correct on exactly this point: a program should do as little as possible and do that little extremely well. That is not nostalgia. It is the only way anybody can reason about what a tool is going to do to their machine. My particular stomachache started last Friday. I am building a new piece of technology, and I needed to open specific applications on macOS, which meant going through [Spotlight](https://support.apple.com/guide/mac-help/mchlp1008/mac). Spotlight is a disaster of a project. It worked in the Jobs era, and over the years it turned into software that never helps with anything. Typing Sla… does not put Slack first. It gives me a thousand results, none of them the one I asked for. It is also a machine for eating resources: its indexing runs for days, gets stuck, and after all that produces nothing useful. My research needs every core the machine has, so waiting for Spotlight to finish whatever it thinks it is doing is time I am not going to get back. So I asked on X whether there were alternatives, some kind people answered with options, and I went to the App Store to find them. Tremendous disappointment. The apps that launch apps on macOS are cheap imitations of Spotlight. Most of them feel like 2007 technology that nobody bothered to update: heavy, buggy, unpleasant to use on a relatively new Mac. And all of them want to be more than a launcher. I want to open applications. I do not want a miniature operating system disguised as a search box. If I want to search my email I go to the email client. If I want to find a file I open Finder. The same is true of every other trick these things offer me, and every one of those tricks is another reason for the app to ask for something it has no business having. Which is exactly what happened. I read one of those Spotlight alternatives 2026 articles that Google puts at the top, went through the comparisons, and picked the one that seemed to have everything I needed and none of what I wanted to avoid. I installed it and was immediately handed a Settings panel full of things to switch on. Fine, I thought, I will switch on nothing and see how far I get. Then I declined Full Disk Access, and the application told me this, verbatim: Limiting Full Disk Access leads to many incorrect bug reports. If you do not want to enable this permission, it is best to quit and uninstall the program. So I uninstalled it. To hell with people who ship that. If your launcher cannot function without reading every file on my machine, the problem is not my permissions, and framing my caution as a support burden for you is an impressive inversion of who is doing whom a favor. I did not want to install a fourth candidate and repeat the ritual, so I decided to write my own. It took me thirty minutes with Codex and GPT-Sol. The thing is fast, it does not ask me for anything, and it does the one job I wanted: it launches applications. [It is on GitHub under an MIT license so anyone can use it or send a patch.](https://github.com/meerita/macos-app-launcher) Thirty minutes is the number worth sitting with. For most of my career, the cost of refusing a piece of bad software was high enough that you swallowed it. You accepted the permission dialog, you accepted the eleven features you would never touch, you accepted that the 2007 codebase would occasionally beachball your Mac, because the alternative was a weekend of work you did not have. That calculation is gone. The tools I now have on this laptop mean the price of saying no to somebody’s bloated launcher is half an hour and a coffee, and once enough people notice that, the only software worth selling in a closed store is software good enough that nobody would bother rebuilding it.