NVIDIA — 7 Open-Source AI Agent Skills

NVIDIA is the author of 7 open-source AI agent skills and MCP servers spanning codex-skill, mcp-server, agent-tool, with a combined 46,310+ GitHub stars. On AgentSkillsHub each is quality-scored (avg 69/100) and security-graded — 7 verified safe.

Top skills by NVIDIA

• NemoClaw — Run agents like Hermes, LangChain Deep Agents, and OpenClaw more securely inside NVIDIA OpenShell with managed inference (22.6k★ · quality 72/100 · security: safe)
• SkillSpector — Security scanner for AI agent skills. Detect vulnerabilities, malicious patterns, security risks, prompt injection, data exfiltration, and s (19.3k★ · quality 76/100 · security: safe)
• skills — Agent Skills for NVIDIA products — install into Claude Code, Codex, and other coding agents to run Physical AI, robotics, simulation, CUDA, (3.5k★ · quality 67/100 · security: safe)
• SkillEvaluator — Multi-tier framework for evaluating AI agent skills with quality gates, semantic overlap detection, synthetic evaluation dataset generation, (533★ · quality 70/100 · security: safe)
• NeMo-Relay — Multi-language agent runtime and library for execution scope management, lifecycle events, and middleware on tool and LLM calls. (190★ · quality 68/100 · security: safe)
• elements — NVIDIA Design System and UI Agent Harness for AI/ML Factories, Robotics, and Autonomous Vehicles (93★ · quality 67/100 · security: safe)
• NeMo-Flow — Multi-language agent runtime for execution scope management, lifecycle events, and middleware on tool and LLM calls. (33★ · quality 65/100 · security: safe)

View NVIDIA on GitHub · Explore audited agent skills