Evidence at a glance #
Based on the linked sources. The author's reported result and independent confirmation are separate checks.
- AI involvement
- Reported models: Claude. Tools: Claude Code, make, Python 3.12, SDL3, FFmpeg, Vulkan. The writeup below describes the work attributed to AI.
- Reported result
- Partial. This is the directory's source-based outcome classification.
- Independent confirmation
- Not reported.
- Human work and limitations
- The project explicitly notes that it is maintained by AI and that the author is not interested in controversy regarding AI-written code. The iOS port is specifically noted as experimental and requires iOS 17.0 or later.
Original sources
What was done #
LAS Recomp is a static recompilation of the 2019 Link's Awakening game for native execution on Windows, macOS, Linux, and experimental iOS. The project relies heavily on AI-generated code and tooling, with human direction and testing.
LAS Recomp is a static recompilation project that enables native execution of The Legend of Zelda: Link's Awakening (2019) on various platforms, including Windows, macOS, Linux, and an experimental iOS port. The project documentation states that it was made almost exclusively with AI, specifically Anthropic's Claude via Claude Code, while being directed, tested, and reviewed by a human. The iOS port, tracked in the ios-port branch, includes touch controls, haptics, and specific adjustments for Apple GPU texture interpolation, such as decoding BC4/BC5 textures at runtime. The project is currently in beta, with known limitations including potential crashes, bugs, and experimental features like dungeon room shuffling. Graphics issues persist, particularly in wide and ultrawide aspect ratios. Users must provide their own legally obtained game files (v1.0.0, v1.0.1, or v1.1.0) to perform the local recompilation. Bug reporting is encouraged via GitHub issues, with logs requested for troubleshooting. Further details are available at https://raw.githubusercontent.com/debeanz/lasrecomp-ios/HEAD/README.md.