Full Vibe-Code (AI generated almost all game code)
Unknown (third party dev
Used REA tool to reverse DOS binary to python and then VS Code with Claude Code and my skill, CPC knowledge files and prebuilt tools to build game in iterations.
ughr.dsk (199.8 KB)
UGH-CPC-Manual-v1.0.pdf (2.4 MB)
It’s fun to play, I don’t remember having tried it before on DOS, cheers!!!
yes, one of my favorite games on DOS, you can play with a friend on same computer, then there are more/different levels (and there are A LOT of levels!)
I did not test it enough, will probably make new version with some improvements but looks like it’s fully playable.
If you have some bug or feature request let me know, I will release the source also at some point, for reference.
And since this turns out to be a successful experiment, maybe I will play around and mod the game / change gameplay or introduce new features… powerups, … this is one of the good things we can do with AI and reversing/recompiling games => we can also be creative, not just “vibe coding”!
So, why would someone not play the far superior PC version over this slow, boring looking, AI-baked CPC version? Is it a fetish? Like “let’s break some games and play them after we make them worse”?
And what exactly is the achievement here? Did the ‘creator’ break any molds when typing the prompts to his LLM? Was he after a 'up to 256 words long" prompt challenge? And what’s next; 128 words? 64?
Welcome to this lovely era of nothing!
Edit: also, I love the ‘didnt play it much but looks playable’ mindset. Whoohoo, the progress of human intellect!
Please keep it civil.
Corrected the tags and adding a ‘wip’ one .
@rexbeng Hello AI hater with no valid arguments yet 
- let me answer:
So, why would someone not play the far superior PC version over this slow, boring looking, AI-baked CPC version? Is it a fetish? Like “let’s break some games and play them after we make them worse”?
Well, I hope you can understand that your question does not make any sense, especially on retro computer forum. You see, these slow machines we all love and although there are faster computers we still want to use from time to time our slow CPC. And, ofc, on slow machine depending on different factors you can get slow games, although I don’t think this port is slow at all! It’s 1x movement for all (in mode 1), with animations, trasparency, music, 2 players, …
And what exactly is the achievement here? Did the ‘creator’ break any molds when typing the prompts to his LLM? Was he after a 'up to 256 words long" prompt challenge? And what’s next; 128 words? 64?
You are missing the point, I do not brag that I achieved anything - I am just sharing experience and a great game port I think people can enjoy. Also, the claim that “you must put effort in order to make things valuable” is pain wrong for many reasons which I can not bother to explain now. But, since you are very interested in the effort I did here is quick summary:
I created a dev container with environment for making the AI able to make CPC games in asm by providing it: mapped knowledge base of different topics, code examples, tricks etc I was able to fetch and organize
In this container I added tools and scripts which can auto update and build from source or download binaries (headless emulator adjusted for even better command line debugging, compiler, disk image tool, gfx image and arkos 3 converters and players, boilerplate code for loader, etc…)
also I added here skill files (runbooks), main being the one for instructions on how to create loop for proper z80 game development with debugging and testing with headless emulator
I used my CPC platform knowledge to write and adjust requirements in order to make the game possible and as good as possible on more restrictive platform like CPC is, so I needed to redesign and decide on resolution, palette, for each sprite, memory layout, decide if colored water is more important then sound, use 4px or 1px movement (speed versus smoothness), …
I made the similar thing making the ready to use dev container for using REA tool and reversed the DOS game to python with the game requirements in text files
(just something from the top of my head, there is probably more…)
All this being said, I would probably love that AI never happened as I think it’s very powerful WEAPON for which the humanity is not ready. But it did happen and it can not be stopped.
Thanks for the reply. However, I will double down. After the commercial era, the point of bringing any game to the CPC has been, up to this point, a developer’s effort to prove his (let’s make it academic, why not) thesis that the computer is entirely capable to execute said game without constraints other than core technical ones, or even enhance the game in some cases using advanced coding knowledge, graphics, sound and even hardware (to overcome those technical constrains).
Feeding other peoples’ intellectual property to an LLM comes nowhere close to the above. You claim you’ve put all that effort to create that dev container and yet, all you did was a conversion of other people’s code. So, while rightfully you acknowledge you “didn’t achieve anything”, you also claim you are sharing experience; are you sure? You just shared what was generated. And then we get the (which I don’t believe I ever brought up in my post, but anyway) “it’s wrong to put effort to make things valuable” notion because, yes, why put effort when you can feed other people’s work to an LLM and generate the result while hav… erm… ing coffee done?
With regards to valid arguments and AI hating, you may take that to that thread because @Gryzor is up to his neck with my ranting and I’d like to save him the effort of having to move posts between threads.
So, really, the question is, what is the point? How about not posting AI generated games? Have you thought of that?
This is what doesn’t make much sense. Porting a game down shows… what, exactly? How well a LLM can work? Oh, wow. Especially when it’s a “didn’t really test it but it looks ok”.
(similarly, I don’t get the porting-up from lesser systems that we see: what’s the point of seeing a game from a machine really inferior to the CPC, running on the latter and looking exactly the same?)
To paraphrase JFK: we do not do these things because they are easy, we do these things because we are odd.