Rars: a Rust RAR implementation, mostly written by LLMs (bitplane.net)

A developer has created "Rars," a Rust implementation of the RAR compression format, primarily utilizing large language models (LLMs) like OpenAI Codex and Claude. Initially, the task of building a full RAR compressor could have taken years, but the project was completed in just five weeks at a cost of around £40 for tokens. Despite resulting in roughly 55,000 lines of code—characterized by some inefficiencies and occasional errors—the tool is functional, marking a significant achievement in automating complex software development using AI. This project is noteworthy for the AI/ML community as it demonstrates the potent capabilities of LLMs in coding and software construction, particularly in reverse engineering and full implementation from scratch. Key takeaways include the effective use of working from specifications, the necessity of extensive testing, and the challenges of performance and development oversight. Although Rars may lag behind established tools like WinRAR in compression efficiency and speed, it represents an innovative foray into leveraging AI for free and open-source software development, ultimately broadening accessibility to the RAR file format. Interested users can install Rars via the command `cargo install rars-cli`.