NURL

NURL is a systems programming language designed for language models, with a syntax optimized for LLM token streams using fixed-arity prefix notation and a one-page grammar. It solves the problem of unreliable code generation by LLMs by reducing ambiguity and error propagation. The language targets developers and engineering teams who build with LLM agents, and is positioned as a self-hosting, LLVM-backed language with deterministic compilation and static borrow checking.

Key features

• Fixed-arity prefix notation
• One-page LL(k≤4) grammar
• Single-owner memory with static borrow checker
• Deterministic compiler with byte-identical output
• LLVM-backed codegen for multiple architectures
• Self-hosting compiler with bootstrap verification
• Algebraic data types and pattern matching
• Closures and higher-order functions
• Generics, traits, match guards, or-patterns
• Try-operator error handling
• Channels with select
• Compile-time const evaluation
• Stackful M:N work-stealing async runtime