NURL
A token-efficient programming language built for language models.
NURL (Neural Unified Representation Language) is a small, LLVM-backed programming language with a regular, prefix-notation grammar optimized for code generated or consumed by large language models. It compiles to native binaries and WebAssembly via LLVM IR, and includes a self-hosting compiler that also exposes an MCP server so AI assistants can build and run NURL programs directly. The language prioritizes token efficiency, local semantics, and deterministic compilation over human-oriented syntax conveniences.
NURL uses a minimal prefix-notation grammar that compiles to LLVM IR, producing native binaries or WebAssembly, with a self-hosting compiler also accessible as a hosted MCP server for AI assistants.
AI/LLM developers and agent builders who generate or consume code programmatically
Background.
- Status
- launched
- Business model
- open-source
Contact
Similar projects.
Editorial take on the space this project sits in — momentum signals, adjacent moves, our call on whether the wedge is real. Get pinged when we publish a new read or when the landscape shifts.
Have a take on this space?
Tell us what you’d build differently, where you think the incumbents miss, or what we’ve gotten wrong about this project. Comments + reactions are coming soon.