← sources

I can't code, and I've released 25 small mobile games in about six months with Claude. Here's how it actually works

source: reddit (reddit:ClaudeAI)published 2026-11-03 13:25ingested 2026-11-04 02:37

retained text

I run a one-person studio making small mobile games, and I've released 25 of them in roughly the last six months, all built with Claude. People sometimes ask how that's possible, so here's an honest breakdown, sorted by stage.



How Claude fits in



I can't code. Every game is plain JavaScript and three.js, wrapped with Capacitor for Android and iOS, and Claude writes all of that code. It also pushes to GitHub and tests the games in Node.js. I never asked for the testing; Claude started doing it on its own. For my number puzzle must be one, all I told Claude was "every board must be solvable, that's a puzzle maker's basic courtesy." The method of building each board backward from a solved state was Claude's idea.



1. Ideas



Most of my ideas come from everyday life.



\- I was in the hospital and had to ride a wheelchair, and I wanted to tear around in it. That became WHEELCHAIR D, a wheelchair racing game.



\- An Uber Eats driver told me waiting for elevators was a pain, and I wanted to control elevators efficiently. That became LOOK EV.



\- I saw a windfoil on TV and wanted to ride one. That became FOIL OF THE DEAD.



\- I was searching for something in my room and wanted to tilt the whole room and slide everything out of the way at once. That became TILT ROOM.



I don't decide the core of a game before I start. I build a rough prototype that just about works. If it looks like it could become something, I keep going. If not, I drop it. I've thrown away several times more ideas than I've actually finished.



2. Production



It varies a lot. Some games have the whole system working in a day, and after that it's just making stage variations. Others take two months of wandering around. SYZYGY RUNNER started out as a pulley-based cargo-lifting puzzle and ended up as a side-scrolling action game.



My role is basically project lead and director. In the past I would have hired a musician for music, an illustrator for art and a coder for code, and managed all of them. Now I give those same jobs to AI.



I usually work on two or more games at once. Since I set my own deadlines, I just pick whichever one I feel like working on. Running an action game and a puzzle game side by side is great for switching gears.



3. Finishing and testing



I don't think there's such a thing as a perfect or complete game. So I ask myself: if I'd had a smartphone in middle school, would I have played this? If the answer is yes, I ship it.



For puzzles, whether a stage can be solved, and for action games, whether a jump can be made and how many tenths of a second of leeway there are: I have Claude calculate those. But in the end, a game is all about how it feels in your hands, so the numbers are just there to back that up.



4. Review and release



I've never failed review on Android. iOS is different. When I submitted several apps close together, they got flagged as spam, and it really looked like nobody had played them. I didn't find any magic trick. I just kept politely asking them to actually look at what was in the games.



Honestly, the place where Claude helped me most was preparing release builds and figuring out the submission process. When there's a correct answer, Claude is incredibly efficient at finding it and building it. Game design has no correct answer, so in the end, that part is still up to me.



Why small games? I just like small things that run on systems. Also, big games need a massive amount of debugging, and that scares me.



5. Marketing



Right now Instagram gets by far the most response. What surprised me is that a single raw run, failures and all, gets a much bigger reaction than a highlight video I edited myself. I'm still learning how X works.



The thing I struggle with most is getting reviews.



6. Keeping it going



My actual working hours haven't changed from when I made things with people. My collaborators are just faster now, so things come out sooner.



Game making is a hobby for me, and I do it within that. I probably have more free time than someone with a regular job. Also, I don't have many friends, so my calendar never fills up with dinners or barbecues.



It's not that I'm trying to pump out a lot of games. It's more that each time one gets done, I want people to play it, and that has just added up.



All 25 games are free to play (with ads) on Google Play, and most are on the App Store too: [https://www.office-canard.com/](https://www.office-canard.com/)



A question for you



Who are you making your games for?



For me, the one who decides whether a game is finished is middle-school me. But the player I'm actually imagining is old-man me. That's probably why I keep making games with simple rules that anyone can play. Big RPGs take a lot of stamina, you know?



Happy to answer any questions.

document fetched 2026-11-04 00:21 · canonical https://reddit.com/r/ClaudeAI/comments/1xmsv5j/i_cant_code_and_ive_released_25_small_mobile