NeoBox: Building a macOS NeoGeo Frontend with AI

Developer NeoBox created a macOS NeoGeo frontend using AI assistance from Claude to implement a native Metal rendering pipeline via libGeoLith, bridging a gap in low-level graphics knowledge. The app allows users to toggle between the GeoLith core or a self-managed MAME installation, targeting an OpenEmu-style experience for SNK classics like Metal Slug or Art of Fighting.

NeoBox: Building a macOS NeoGeo Frontend with AI Claude /en/tags/claude/ is surprisingly capable at handling graphics rendering pipelines, which is exactly how NeoBox ended up supporting a native Metal pipeline via libGeoLith. Most of the app is standard AppKit code I wrote by hand, but the low-level rendering stuff was a complete blind spot for me. Using AI to bridge that gap allowed me to move beyond a simple MAME frontend and actually implement something with better performance on macOS. For anyone curious about the technical split in this project: Manual Code: The entire native macOS UI and the tedious process of ROM compatibility testing. AI-Assisted: The website and the integration of libGeoLith/Metal. Without an LLM, I would've been stuck in documentation hell trying to figure out the rendering pipeline. The app currently lets you toggle between the GeoLith core or a self-managed MAME installation. It's basically designed for people who want an OpenEmu-style experience but specifically for SNK classics like Metal Slug or Art of Fighting. If you're looking for a practical tutorial on how to integrate C++ libraries into a Swift/AppKit project using AI, this is a great real-world example of using an LLM agent to handle the "scary" math and graphics code while the human handles the architecture and UX. https://apps.apple.com/us/app/neobox-neo-geo-player/id6769912877 Next Unlayer: A Complete Guide to Embeddable Builders → /en/threads/2297/ All Replies （3） A D F Claude's actually been a lifesaver for my Swift projects lately, saves so much boilerplate. 0