The ARO Programming Language was fully implemented with artificial intelligence.
This entire project—including the parser, compiler, runtime, documentation, website, CI/CD pipeline, and this disclaimer—was created through collaboration between a human and Claude, an AI assistant developed by Anthropic.
- All Swift source code was generated by Claude
- The lexer, parser, semantic analyzer, and runtime were AI-written
- All 27 language proposals were drafted by AI
- Test suites were AI-generated
- Language guides, tutorials, and API references
- The website (HTML/CSS)
- This README and all markdown files
- GitHub Actions CI/CD pipelines
- Dockerfile and deployment scripts
- Build system configuration
A human provided:
- High-level direction and requirements
- Design decisions and feedback
- Review and iteration guidance
- The vision for a feature-driven, AI-friendly programming language
We believe in transparency about AI's role in software development. ARO itself is designed to be AI-friendly—a language where the gap between human intent and machine execution is minimal. It's fitting that ARO was built by AI.
This project demonstrates:
- AI can build complete, working programming languages
- Human-AI collaboration produces functional software
- The line between "human-written" and "AI-written" code is blurring
While AI-generated, the code has been:
- Compiled and tested (262 passing tests)
- Iteratively refined through feedback
- Designed following established patterns
However, AI-generated code may contain:
- Subtle bugs not caught by tests
- Suboptimal implementations
- Inconsistencies across the codebase
Use at your own discretion. Contributions and bug reports are welcome.
This AI-generated code is released under the same license as the project. AI-generated content can be freely used, modified, and distributed.
Built with Claude Code. Obviously.