Amiga Unix Revival Project Uses Generative AI to Restore Legacy OS

The project introduces a modern toolchain, a package manager, and drivers for previously unsupported hardware. Installation is handled via floppy or CD-ROM, supporting both interactive and unattended setups on real hardware or emulators like WinUAE and FS-UAE. The apkg package manager allows users to install tools—starting with grep, gzip, less, patch, and zlib—directly from a dedicated repository.

A distinctive aspect of the development process is the use of generative AI to assist in reverse engineering. Because source code is unavailable for many components, LLM models are used to read old kernels in binary format to help write drivers and technical notes. These AI-generated contributions are reviewed and tested by community members on real hardware to ensure stability.

The project maintains an open documentation system called the "grimoire," which uses Markdown to record both human-authored and AI-assisted work. Each entry is assigned a confidence tag—ranging from "verified" to "unverified"—to indicate the level of certainty regarding the technical information provided.

PLUS ULTRAby Amenoyomi

Because original source code is unavailable, generative AI is used to analyze the old kernels in binary format. This process allows the models to assist in drafting new drivers and producing technical notes that would otherwise require manual reverse engineering of the machine code.

To ensure these AI-generated contributions are reliable, the project employs a strict human-centric verification loop. Community members set the overall technical direction and review every change made to the code. The final validation occurs on real hardware, where the success of a driver is determined by a binary outcome: the system either boots or it does not.