• Hello Guest! We're hosting a challenge to welcome vintage Intel macs to the MLA during the month of July! See this thread for more information.
  • We've made some quality of life improvements to the Trading Post. More info here.

My First AI Program Sourcecode - KNPR

olePigeon

68040
Below is the source code to my first (and only) app created using a LLM (you can read the original development thread here.) My work changed our AI stuff for COPA compliance and it no longer remembers anything past the current session. So after a month of trial and error, I've lost all the progress with the LLM agent. I have to retrain the stupid thing every time, and that takes hours every time with a bunch of trial and error because it constantly ignores my instructions and sources code outside of its set boundaries. So I've had to abandon (for now) my plans to implement a full Editor. :( I guess this is good news for people worried about "Vibe Coders."

I'm also not interested in buying AI anything. I don't know if there are any free non-networked coding agents. Maybe Visual Studio will get free local AI stuff eventually, at which point maybe I can continue working on it. So if there are any major bugs in it, I can't fix it at the moment.

Since I don't know how to program, I'd be curious if the AI generated legible code.

So here's the final KNP Reader app and source code. It can read C&N files, so you can use it as a regular C&N reader. My slightly updated file format allows for some extras like color support and linking Patches (for running kracks and patches on newly installed apps.) It also has better sorting than C&N and I implemented a Find feature for those huge databases.

KNP Reader.gif
 

Attachments

Maybe Visual Studio will get free local AI stuff eventually, at which point maybe I can continue working on it.
Pretty cool that AI can (kinda) write classic mac programs now. If you have a fast enough computer, you can connect local AI models to Visual Studio Code. You're going to need either an Apple Silicon Mac with probably 32GB minimum of unified memory (RAM), or a PC with about the same 24GB minimum of GPU VRAM. But if you set up a local llama.cpp or ollama server, you can pretty easily patch it into your VS Code, and it works about the same as GitHub Copilot but completely offline and free. There's guides online about this, or you can always ask AI! :)
 
Back
Top