Overview
- NVIDIA has released a large set of open-source “physical AI” skills inside the NVIDIA Agent Toolkit that make its Omniverse, Isaac, Metropolis, Alpamayo, Cosmos and Jetson components callable by coding agents.
- The skills break complex workflows into repeatable agent tasks so agents can automate data generation, run simulations, train and evaluate models, and prepare systems for real-world deployment.
- NVIDIA bundled governance features including the NemoClaw blueprint and the NVIDIA OpenShell runtime to enforce policy-based security and privacy when agents run locally or in the cloud.
- The toolkit and published skills are available now on GitHub and skills.sh for use with any coding agent, and vendors and teams are already using them to accelerate robotics, autonomous vehicle and factory workflows.
- The move builds on years of NVIDIA work in simulation, vision and edge AI and could speed development cycles, lower integration friction for developers, and shift more testing and tuning into automated, agent-driven workflows.