github.devotd.
NewRobotics

syswonder/robonix v1.0.0 — RTDL-based robot agent OS

syswonder/robonix is a robot agent operating system written mainly in Rust that turns natural-language task instructions into RTDL intermediate-representation programs, validates them, and runs them; the latest release is v1.0.0.

Repositorysyswonder/robonix ↗

syswonder/robonix is a robot agent operating system written mainly in Rust that turns natural-language task instructions into RTDL intermediate-representation programs, validates them, and runs them; the latest release is v1.0.0.

GitHub description: The Agentic Operating System for Robots

stars
★ 402
Forks
57
Language
Rust
License
MulanPSL-2.0
Latest release
v1.0.0
Read original source ↗View on GitHub ↗

syswonder/robonix is an agent operating system for robots. The repository describes itself as "The Agentic Operating System for Robots" and is written mainly in Rust. Its aim is to use LLMs and VLMs to convert natural-language task instructions into programs a robot can execute, and then run those programs. The repository was created on September 25, 2025, and the latest release is v1.0.0.

The conversion result is a program in RTDL, a general-purpose intermediate language for robots. The idea is to compile natural language once into an intermediate representation and execute that. Software capabilities are separated from hardware through four layers — Task, Skill, Service, and Primitive — and versioned contracts, so implementations can be replaced or reused across several robot platforms. Programs are validated before execution, and specified results can be checked once a run finishes.

Hardware is generalized through a layered capability abstraction and IDL interfaces. The license is MulanPSL-2.0 and the repository has 402 stars. It is a hybrid codebase that mixes Python libraries with Rust code, and it even carries a 1.1MB lock file, which points to real development activity.

However, the general-purpose claim is stated without revealing which robots are actually supported or how many, so it is hard to verify. Neither the changes in v1.0.0 nor any real deployment cases are given in this record. On a 10-point scale the scores are novelty 7.0, tech 7.0, quality 6.0, and momentum 0.5, and the low momentum puts it close to an early-stage project. Its generality is too early to judge until more measured data accumulates.