Show HN: A 100% vibecoded programming language interpreter (github.com)

A new project called Vibelang demonstrates an end-to-end programming language interpreter that was written entirely by an AI. The repository (github.com/mrunix00/vibelang) includes build instructions: clone with --recursive, run make to build, optional make test, and run the interpreter with ./build/main <file.vibe>. The repo also contains a syntax specification so you can inspect the language grammar and see what the generated code implements. In short, this is a runnable, AI-produced interpreter rather than just snippets or scaffolding. The release is significant as a practical proof-of-concept that current AI tooling can produce complete system software components — a real interpreter with build scripts and tests — which accelerates language prototyping and automation in tooling. For the AI/ML community it highlights both opportunity and caution: such artifacts speed developer workflows and experimentation, but require rigorous validation (security auditing, correctness testing, dependency checks) because model-generated code can introduce subtle bugs or unsafe constructions. Technically, the artifact is an interpreter (not a compiler), includes a formal spec to verify behavior, and provides tests to aid reproducibility; it’s a useful case study for automated code generation, reproducible engineering with AI, and the limits of trusting end-to-end model-authored systems.