How much of F-Droid is LLM generated?

I love F-Droid. I love what F-Droid stands for and I like the freedom that it gives its users. As a FOSS app maintainer I also have nothing but good things to say about the people behind the project. Maybe they’re even a bit too nice, considering how many times repro has failed due to me forgetting to commit before building 😅.But my god is it hard to find human-written software now.IntroThe problemSometimes I open F-Droid just to browse. You know, maybe I’ll find an app that solves a problem I didn’t know I had, or maybe I’ll find a better alternative to something I already use.And one day, while browsing, I noticed an app with an obvious and ugly AI generated icon. It’s not on the list and I will not shame it, but it got me thinking. How much of F-Droid is AI?As someone who does programming for fun and is only a student, untarnished by honest work, I get my knowledge of what the coding world is like mostly from clickbaity YT videos and Reddit posts of CS professionals. They either describe LLMs as god reincarnate or as glorified autocomplete.Both are, obviously, wrong, but that’s not really helpful in determining what I want to know. What is the actual state of programming nowadays? Whenever I use FOSS software, how likely it is that it has been vibe-coded by a rando in an afternoon?How to know if an app is vibe-coded?That’s the trick—you can’t. Text just doesn’t carry enough meta information for any kind of assessments to be even close to accurate. However, just as that em-dash I used in the first sentence probably triggered an alarm in your brain, there are signs.While that does make the task at hand sound fickle and dependent on happenstance, the signs, especially concerning LLM code repositories, are never too hard to find.You see, the main allure of LLMs is that they allow the developer to be more lazy. That’s kind of the whole point! You just prompt, sit back and relax. So it should not surprise you to hear that this attitude is then reflected in everything the vibe-coder touches.Why write a README from scratch? Lol, just let the LLM do it.Do code-review? Naw, just let the LLM review its own changes and then also give it access to the repo so you don’t even need to press the commit button.If a project was concerned with looking legitimate, it would be trivial to do such things by hand. But it’s low effort all the way down.My biasesThis part can be skipped if you don’t care about my stance on LLMs, but I need to make my position clear to avoid contributing to the circlejerk too much.To begin with, I’d like to acknowledge that LLMs are incredibly useful and capable. As 2026 has progressed, this has become more and more visible, but people being able to one-shot medium scale games and software in half an hour is ridiculously impressive, even if the end product is usually not very good.I also am very much in favor of software getting faster and more secure. The story of the Linux kernel development has shown that LLMs are capable of finding and sometimes even solving many types of code issues.With the pleasantries out of the way though, I have to admit I really hate LLMs and what they have done to programming, related engineering fields, and society as a whole.Their mere existence makes educating yourself and going on fun side projects much less rewarding. Like yeah, I did something, but with an LLM I could have done this in a quarter of the time. And when you do take the black pill and vibe-code, it’s even worse. It’s not like you did anything. The machine did that.Then there’s the atrophying effects on human brains, their unfathomable capability for serving plausibly sounding misinformation and the climate disaster that we’re just kinda ignoring. All very fun things to think about.Criteria / The experimentAs mentioned before, there’s no way to effectively detect slop, so I propose a rough 3 tier system based on the aesthetics of the repo:Mostly AIThis is mostly for projects that have significant LLM smells and means I expect >50% of the code is LLM authored. Any kind of agentic infrastructure automatically lands an app in this tier as I do not believe it is possible to use AI responsibly from within a coding harness.Hard to say / Mostly human / OtherOccasional LLM commits either by maintainers or contributors, but mostly looks human. May have an LLM policy which permits certain uses. This tag means I expect <50% of the code is LLM authored.No signs of AICould not find anything suspicious/Has a strict LLM policy.Note that my rating will still be quite superficial and the tiers quite loose. I did not build a “slop detector” or anything like that, so my decisions are based on looking at recent commits and their content along with the project’s branding.Also note that since I have no way of knowing for sure, there may be errors. I still believe most of my findings to be accurate however.I will also not look at the history of the app. If it has existed since 2014 but recent commits are LLM authored it will be categorized as “mostly AI”The apps chosen were the batch of updates pushed to F-Droid on September 12, 2026. That amounted to 102 apps, which was quite a lot of work for me to go through.😅If you just want the results, feel free to skip to the resultsAppsAmberDescription: Nostr event signer for AndroidRepository: https://github.com/greenart7c3/AmberRating: Mostly AIJustification: All recent commits were done with LLM, PRs accepted from agents, Claude Code and Codex infrastructure present.Aria for MisskeyDescription: Dive into the interplanetary microblogging platform 🚀Repository: https://github.com/poppingmoon/ariaRating: No signs of AIJustification: This is a hard one as commit naming and structure felt a bit suspicious, but I did not see anything else and decided to err on the side of cautionAtmo EngineDescription: Animated wallpapers with Atmosphere, Glass, Canvas Sketch and playlists.Repository: https://github.com/saad-khan-rind/nosatmosphereeffectRating: No signs of AIJustification: Heavy emoji usage in the README, but otherwise nothing looks suspicious.Aves LibreDescription: Gallery and metadata explorerRepository: https://github.com/deckerst/avesRating: No signs of AIJustification: Nothing looked suspiciousBalanceDescription: Private offline bank balance dashboardRepository: https://github.com/AshkanRafiee/balanceRating: Mostly AIJustification: Code and commits exhibit various AI smells.Baly Groceries TrackerDescription: Track your groceries, without worrying about running out.Repository: https://github.com/rw-account/baly_groceries_trackerRating: Mostly AIJustification: Code and commits exhibit various AI smells.Bati: Fitness RPGDescription: Turn workouts into quests, boss fights and a village built by your training.Repository: https://github.com/Guiforge/batiRating: Mostly AIJustification: AI disclosure in README (Thanks for making it easy :D)BayesianBahnDescription: Empirical arrival-time distributions for Deutsche Bahn trainsRepository: https://github.com/DerWeh/BayesianBahnRating: Mostly AIJustification: All commits are Claude co-authored. Though it’s really funny how Deutsche Bahn sucks so hard it has enticed people to vibe-code unofficial timetables.BeatBridge: Bluetooth MusicDescription: Auto-play music when your Bluetooth car, headphones, or speakers connect.Repository: https://github.com/brandonp2412/BeatBridgeRating: Mostly AIJustification: Includes Claude Code contributions, quite a lot of the commits look to be at least AIBinary EyeDescription: QR code and barcode scanner with no adsRepository: https://github.com/markusfisch/BinaryEyeRating: No signs of AIJustification: Nothing suspiciousBlockDrop: Block PuzzleDescription: Free and open source block-stacking puzzle game. Arrange falling shapes to clearRepository: https://github.com/brandonp2412/BlockDropRating: Mostly AIJustification: Has infrastructure for LLMs, Claude co-authored commits.Braincup - Brain TrainingDescription: Train your focus, memory and math skills.Repository: https://github.com/SimonSchubert/BraincupRating: Mostly AIJustification: All recent commits are Claude co-authoredBVDDescription: bilibili video resource downloaderRepository: https://github.com/KafuuNeko/BiliDownloadRating: Mostly AIJustification: All recent commits look LLM authoredCaptureCapDescription: Record and Stream Audio and/or ScreenRepository: https://github.com/yepgoryo/CaptureCapRating: Mostly AIJustification: Has an AI disclosure listing files written with LLM helpCasio G-Shock Smart SyncDescription: Add smart functions to your Casio square BT G-Shock Watch.Repository: https://github.com/izivkov/CasioGShockSmartSyncRating: Mostly AIJustification: AI artifacts such as planning .md files pushed to repo.Chompass - Calorie TrackerDescription: Ad-free AI calorie tracker. Snap, speak, or type meals. Open source.Repository: https://codeberg.org/fitguy/chompassRating: Mostly AIJustification: Recent commits look AI authored. Note that this repository is hosted on Codeberg and likely violates their terms of use.DeltaSyncDescription: Encrypted KeePass sync — requires your own self-hosted server.Repository: https://gitlab.com/Star95/keepass-deltasyncRating: Mostly AIJustification: All recent commits co-authored by Claude Opus 5DuressKeyboardDescription: Keyboard that reacts to triggers and performs security actionsRepository: https://github.com/pofesk0/lastcodeduresskeyboardRating: ???Justification: This may be the most confusing repo I’ve ever seen lol. Has been in development since November of 2025, but the owner doesn’t seem to know how to use git??? All the commits have seemingly been done by copy pasting the change code by hand through the GitHub web file editor. Quite bizarre!eQuranDescription: Read, listen, choose from 5 reciters, and track your Quran journey.Repository: https://github.com/ya27hw/equran_appRating: Mostly AIJustification: Has infrastructure for agents; most recent commit looks LLM generated.evcc - solar chargingDescription: charge your