NVIDIA has released a major collection of open-source physical AI skills and developer tools through its Agent Toolkit, giving software developers pre-built, agent-executable instructions for building robotics, autonomous vehicle, vision AI and industrial digital twin systems without starting from scratch.
What the Agent Toolkit release includes
NVIDIA made the announcement at GTC Taipei on June 1, 2026, positioning the open-source release as a way to help developers turn complex physical AI workflows into agent-executable tasks, reducing the costs, time and complexity of building those pipelines at scale.
The physical AI skills, available as part of NVIDIA Agent Toolkit, let agents use NVIDIA libraries, models and frameworks to speed the data generation, simulation, training, evaluation and deployment pipelines behind robots, AVs, factories and labs.
The collection spans NVIDIA Omniverse, Cosmos, Alpamayo and Metropolis, covering robotics, autonomous vehicles, vision AI and industrial digital twins.
How the tools work for developers
NVIDIA designed the new skills to turn physical AI development processes into repeatable instructions that coding agents can follow, including which tools to call, what outputs to produce and how developers can validate results.
Developers can build and deploy autonomous agents using these skills together with the NVIDIA NemoClaw blueprint and the NVIDIA OpenShell runtime, which provides policy-based security and privacy governance on local or cloud hardware.
The toolkit matters to developers because it abstracts the most time-consuming parts of a physical AI pipeline. Rather than hand-coding every data generation or simulation step, a coding agent can now call a pre-validated skill and follow a structured path to a deployable result.
That mirrors the shift already well underway in software-only AI, where NVIDIA’s Nemotron open models have attracted enterprise partners who cite inference cost advantages over proprietary alternatives.
Industry partners already using the stack
Industry leaders including Agile Robots, Cadence, Dassault Systèmes, Delta Electronics, Foxconn, Pegatron, PTC, Siemens, Synopsys and TSMC are using NVIDIA physical AI tools to accelerate physical AI development.
As part of its Autonomous Fab 2030 roadmap, SK hynix is implementing semiconductor fab digital twins using NVIDIA Omniverse, and collaborating with NVIDIA and SK Telecom to validate NVIDIA Agent Toolkit for manufacturing-specific physical AI.
Robotics companies including 1x, Agile Robots, Agility, FieldAI, Hexagon Robotics, NEURA Robotics, Skild AI and Universal Robots are among those using NVIDIA’s agent-ready physical AI stack to accelerate robotics development from data generation to deployment.
Why the open-source move matters
The new physical AI skills are designed to turn complex physical AI training, evaluation and deployment workflows into repeatable, optimized and agent-executable instructions. That design philosophy borrows from software engineering best practices and applies them to the harder problem of making machines move and perceive in the real world.
Jensen Huang, NVIDIA’s founder and CEO, said: “AI agents are revolutionizing software development, and that shift is now coming to physical AI, extending into the systems that will transform transportation, manufacturing, healthcare and robotics.”
For developers already building with NVIDIA infrastructure, the toolkit is a direct productivity lever. Developers can now use pre-built skills to automate data generation, simulation and evaluation pipelines, cutting costs and development time.
That efficiency argument is consistent with how NVIDIA has positioned open-source releases broadly: lowering the barrier to entry for teams that want access to state-of-the-art models and frameworks. The company’s recent DGX GB300 donation to the Naval Postgraduate School reflected a similar logic applied to hardware access rather than software.
Full documentation and the open-source skills are available through GitHub and skills.sh, with preconfigured launch environments also available through NVIDIA Brev.
