🎯Skills353
An official NVIDIA-verified agent skill that teaches AI agents how to use NVIDIA CUDA-X libraries and platform tools correctly — part of a catalog of portable instruction sets mirrored daily from product repos.
An NVIDIA-verified skill that guides AI agents through deploying, configuring, troubleshooting, and managing retrieval augmented generation (RAG) pipelines using Docker Compose or Helm, based on NVIDIA's RAG AI Blueprint.
Guides AI agents through NVIDIA cuOpt developer tools and workflows for building GPU-accelerated optimization applications including vehicle routing, linear programming, and quadratic programming.
An NVIDIA-verified skill that provides agentic guidance for developing applications with NVIDIA DeepStream, a platform for building AI-powered video analytics and intelligent video applications on NVIDIA GPUs.
An NVIDIA-verified skill that teaches AI agents how to work with agent skills within NemoClaw, NVIDIA's secure sandboxing platform that runs OpenClaw inside NVIDIA OpenShell with managed inference, policy management, and monitoring capabilities.
Sends deep research queries to a locally running NVIDIA AI-Q Blueprint server, polling asynchronous jobs and returning reports with citations and source URLs intact.
An official NVIDIA-verified skill for AI agents, providing portable instructions for using NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly, installable via the skills CLI.
Official NVIDIA-verified skill that teaches AI agents how to use NVIDIA cuDF for GPU-accelerated DataFrames, including pandas acceleration, dask-cuDF, ETL workflows, joins, groupby, CSV/Parquet I/O, nullable semantics, and multi-GPU DataFrame workloads. Part of the NVIDIA Agent Skills catalog with 110+ verified skills across 24 products.
Guides AI agents through importing and configuring vision models in NVIDIA DeepStream for video analytics and intelligent streaming applications.
An official NVIDIA-verified skill for AI agents, providing portable instructions for using NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly, installable via the skills CLI.
An NVIDIA-verified skill that provides a getting started guide for NemoClaw users, helping AI agents walk users through initial setup of NVIDIA's secure agent sandboxing platform for running OpenClaw inside NVIDIA OpenShell.
An NVIDIA-verified skill that guides AI agents through configuring security settings for NemoClaw, NVIDIA's secure agent sandboxing platform for running OpenClaw inside NVIDIA OpenShell with managed inference and policy management.
An official NVIDIA-verified skill for AI agents, providing portable instructions for using NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly, installable via the skills CLI.
Guides AI agents through configuring managed inference settings for NemoClaw, NVIDIA's secure agent sandboxing platform that runs OpenClaw inside NVIDIA OpenShell.
An official NVIDIA-verified skill for using the cuOpt server API in Python, enabling AI agents to solve GPU-accelerated optimization problems including vehicle routing, linear programming, and quadratic programming through NVIDIA's cuOpt platform.
NVIDIA-verified skill for managing user sandboxes in NemoClaw, which provides secure agent sandboxing for running OpenClaw inside NVIDIA OpenShell. Covers sandbox lifecycle operations including creation, monitoring, and policy management for secure agent execution environments.
An NVIDIA-verified skill that teaches AI agents user rules and conventions for cuOpt, NVIDIA's GPU-accelerated optimization library for vehicle routing, linear programming, and quadratic programming problems.
Guides AI agents through NVIDIA cuOpt's C API for GPU-accelerated numerical optimization, covering linear programming and quadratic programming tasks.
Guides AI agents through managing security and access policies in NemoClaw, NVIDIA's secure agent sandboxing platform for running OpenClaw inside NVIDIA OpenShell.
Handles installation, deployment, validation, troubleshooting, and shutdown of NVIDIA AI-Q Blueprint infrastructure via Docker Compose, Python/uv, Node.js, or Kubernetes/Helm.
Provides an introductory overview and onboarding guide for NemoClaw, NVIDIA's secure agent sandboxing platform that runs OpenClaw inside NVIDIA OpenShell with managed inference and monitoring capabilities.
Guides AI agents through deploying and managing NVIDIA cuOpt server instances for GPU-accelerated optimization, covering common server setup and configuration patterns.
Inspects NemoClaw sandbox health, traces agent behavior, and diagnoses problems using the OpenShell CLI status, logs, and TUI tools together.
NVIDIA-verified skill for USD performance tuning in Omniverse, part of the Physical AI suite covering optimization of USD scenes and Kit-based application performance.
An NVIDIA agent skill that provides an onboarding guide for CUDA Quantum, covering installation, test programs, GPU simulation, QPU hardware access, and quantum applications.
Architecture reference for the NemoClaw plugin and blueprint system that orchestrates the OpenClaw sandbox, covering plugin structure, blueprint design, CLI commands, network policies, and troubleshooting.
NVIDIA Omniverse skill for CAD-to-SimReady asset conversion workflows, enabling automated pipeline setup for converting CAD models into simulation-ready USD assets.
An NVIDIA-verified skill that guides AI agents through remote deployment workflows for NemoClaw, enabling users to deploy secure agent sandboxes with managed inference on remote infrastructure via NVIDIA OpenShell.
An official NVIDIA-verified skill that teaches AI agents how to use NVIDIA DALI (Data Loading Library) in dynamic mode for GPU-accelerated data loading and preprocessing pipelines. It is part of NVIDIA's catalog of agent skills covering CUDA-X libraries, AI Blueprints, and platform tools, maintained in product repos and mirrored to this central catalog.
Generates or updates NVIDIA governance skill cards for existing agent skills by gathering source signals, building a grounded JSON context, and rendering a deterministic markdown card for legal/safety review.
NVIDIA-verified skill for Omniverse realtime viewer orchestration, part of the Physical AI suite for USD/Kit workflows in simulation and synthetic data generation environments.
NVIDIA RAG Blueprint skill for evaluating retrieval augmented generation pipelines, covering deployment validation, performance assessment, and troubleshooting of RAG systems with Docker Compose or Helm.
NVIDIA cuOpt skill for GPU-accelerated optimization covering vehicle routing, linear programming, quadratic programming, server deployment, and developer tools.
NVIDIA Physical AI skill for neural reconstruction (NuRec/NRE), supporting infrastructure setup for synthetic data generation, resilient scaling, and 3D scene reconstruction workflows.
A CLI tool that indexes folders of documents—PDFs, images, Office files, HTML, audio, and video—into LanceDB and performs vector search, verbatim quoting, and corpus-level aggregation across them.
NVIDIA RAG Blueprint skill for performance optimization and monitoring of retrieval augmented generation pipelines, supporting deployment tuning with Docker Compose or Helm.
NVIDIA cuOpt skill for formulating GPU-accelerated numerical optimization problems including linear programming, quadratic programming, and vehicle routing problem setup.
NVIDIA Physical AI skill for infrastructure setup supporting synthetic data generation and resilient scaling of Physical AI development workloads.
NVIDIA-verified skill for formulating vehicle routing problems with NVIDIA cuOpt, a GPU-accelerated optimization library supporting VRP, TSP, linear programming, and quadratic programming.
End-to-end Nemotron model development skill for customization, evaluation, and deployment pipelines using the NVIDIA AI stack.
NVIDIA-verified skill for installing cuPyNumeric, which enables NumPy and SciPy workloads to run on multi-node multi-GPU systems without code changes.
An NVIDIA-verified skill that guides AI agents through developing training recipes with NeMo AutoModel, enabling PyTorch-native distributed training configurations for LLMs and VLMs with customizable launchers and validation workflows.
PyTorch-native distributed training skill for LLMs and VLMs using NVIDIA NeMo AutoModel, with Hugging Face model support, training recipes, multi-GPU launchers, and validation workflows.
NVIDIA NeMo AutoModel skill for configuring launcher settings in PyTorch-native distributed training workflows for LLMs and VLMs, with Hugging Face support, recipes, and validation workflows.
Guides AI agents through building synthetic datasets using NVIDIA's Data Designer library. Supports interactive and autopilot modes with declarative pipeline generation, category/Jinja2 samplers, and seed dataset integration.
An NVIDIA-verified skill that guides AI agents through onboarding models with NeMo AutoModel, which provides PyTorch-native distributed training for LLMs and VLMs with Hugging Face support, recipes, launchers, and validation workflows.
An NVIDIA-verified skill that guides AI agents through starting router modes for NVIDIA Dynamo deployment on Kubernetes, part of the Dynamo skill set covering recipe deployment, interconnect validation, and day-2 troubleshooting.
An NVIDIA Physical AI skill for video data augmentation in synthetic data generation pipelines, supporting Omniverse USD/Kit workflows for physical AI development.
Assists with planning, debugging, tuning, evaluating, exporting, and deploying public Nemotron embedding and reranking retrieval recipes, preferring the current source checkout over cached knowledge.
Pre-migration readiness assessor that inspects NumPy code to determine GPU scalability via cuPyNumeric, identifying blockers like unsupported APIs, scalar synchronization, and host round-trips, then providing a verdict of READY, LIGHT REFACTOR, SIGNIFICANT REFACTOR, or NOT RECOMMENDED.
An NVIDIA-verified skill that guides AI agents through picking and deploying recipes for NVIDIA Dynamo on Kubernetes, enabling automated deployment workflows for distributed inference serving infrastructure.
An NVIDIA-verified skill that generates labeled synthetic training data for physical-AI inspection and perception models, enabling defect detection workflows without real-world image collection.
An NVIDIA-verified skill that guides AI agents through validating disaggregated NIXL, UCX, and NCCL interconnects for NVIDIA Dynamo deployments on Kubernetes, ensuring proper network connectivity for distributed inference.
Loads sharded on-disk datasets (sharded .npy, Parquet/Arrow, raw binary, sharded HDF5) into distributed cuPyNumeric ndarrays via manual partition and leaf @task launches with CPU/OMP/GPU variants.
Guides users through the PhysicsNeMo repository to find the right models, datapipes, and examples for SciML and AI4Science tasks like surrogates, forecasting, downscaling, and physics-informed simulations.
An NVIDIA-verified skill that guides AI agents through triaging day-2 failures and troubleshooting NVIDIA Dynamo deployments on Kubernetes, covering interconnect issues, router problems, and deployment diagnostics.
An NVIDIA NeMo Platform skill for designing and managing training data pipelines using the NeMo unified CLI, Python SDK, and web UI.
Provides techniques for resolving GPU OOM errors in Megatron Bridge training, including expandable memory segments, parallelism resizing, activation recompute strategies, and CPU offloading constraints.
Reads and writes large cuPyNumeric arrays to HDF5 files using Legate's parallel, distributed HDF5 I/O (to_file, from_file, from_file_batched), with optional GPUDirect Storage acceleration.
Playbook for launching, monitoring, stopping, and debugging NeMo-RL training recipes on Kubernetes clusters via the nrl-k8s CLI, supporting both ephemeral and long-lived RayCluster modes.
Guides validation and usage of CUDA graph capture in Megatron Bridge to reduce host-driver overhead, covering both local full-iteration graphs and Transformer Engine scoped graphs for attention, MLP, and MoE modules.
Summarizes recorded videos using NVIDIA's LVS summarization microservice with human-in-the-loop gating and a VLM fallback, providing end-to-end workflows via curl commands.
Converts single-node training scripts to multi-node Slurm sbatch jobs for Megatron Bridge, covering srun-native launches, Enroot container setup, NCCL timeout debugging, and MoE model OOM sizing.
Operational guide for enabling hierarchical context parallelism in Megatron Bridge, using nested context-parallel process groups with a2a+p2p communication to scale CP beyond KV heads.
Guides configuration of CPU offloading in Megatron Bridge, covering both layer-level activation offloading and fractional optimizer state offloading via HybridDeviceOptimizer to reduce GPU memory usage.
Guides installing Earth2Studio via uv or pip, selecting the correct model extras, and configuring the Python environment for weather and climate AI inference.
Operational guide for choosing and combining TP, DP, PP, CP, and EP parallelism strategies in Megatron Bridge, with sizing rules by model size and hardware topology for both dense and MoE models.
Guides running Megatron-LM and Megatron Bridge training with mock or real data, covering correlation testing between MLM and Bridge loss curves, available recipes, and multi-GPU launch configuration.
Provides guidance for training MoE models at long sequence lengths in Megatron Bridge, covering context parallelism sizing, selective recompute, dispatcher choices, and patterns from DSV3 and Qwen3 experiments.
Guides selective and full activation recompute in Megatron Bridge to reduce GPU memory usage, with a prioritized checklist from expandable segments through core_attn, layernorm, and mlp recompute modules.
An NVIDIA NeMo Platform skill for evaluating AI models using the NeMo unified CLI, Python SDK, and web UI for model assessment and benchmarking.
Indexes all shipped Megatron Bridge recipes and recommends the best starting training configuration based on model name/size, GPU count and type, and training goal (pretrain, SFT, or PEFT).
Practical guidance for training MoE vision-language models in Megatron Bridge, comparing FSDP and 3D-parallel approaches with lessons from Qwen3-VL and other multimodal experiments.
Deploys and operates the RTVI-CV-3D multi-view 3D tracking stack with per-camera DeepStream perception and BEV Fusion, supporting sample data, custom videos, and RTSP flows with automatic calibration chaining.
Stage 1 of NVIDIA's Clinical ASR Flywheel that bootstraps a clinical speech recognition cycle, including NVCF/MW disclosure, API key checks, dependency installation, and TTS+ASR smoke tests.
Helps choose the right MoE token dispatcher (alltoall, DeepEP, or HybridEP) for Megatron Bridge based on hardware platform, EP degree, and optimization stage, with patterns from DSV3, Qwen3, and VLM work.
Enables MoE expert-parallel communication overlap in Megatron Bridge to hide dispatch/combine latency, supporting alltoall, DeepEP, and HybridEP flex dispatcher backends with optional delayed weight-gradient computation.
Asks visual questions about recorded video clips using the VLM through NVIDIA's VSS agent, providing fresh model inference on video frames when existing summaries or metadata are insufficient.
Official NVIDIA-verified skill for tile-based GPU programming that guides AI agents through adding new CuTile kernels, performing cross-framework conversion, and optimizing performance. Part of the NVIDIA Agent Skills catalog with 110+ verified skills across 24 products, each signed and benchmarked.
An NVIDIA Earth2Studio skill for running deterministic weather forecasts using a deep-learning framework for building and deploying AI weather and climate workflows.
Systematic MoE training optimization workflow for Megatron Bridge based on the Three Walls framework (memory, communication, compute), covering parallel folding, recompute strategy, dispatcher choice, and CUDA-graph bring-up.
Configures MoE expert-parallel communication overlap in Megatron Bridge to hide token dispatch/combine latency, covering flex dispatcher backends and expert weight-gradient scheduling for throughput tuning.
Deploys the VSS video-analytics-api REST service standalone with custom config, data-log binds, and Elasticsearch/Kafka connectivity for NVIDIA's Video Search and Summarization blueprint.
Helps users find the right Earth2Studio models, data sources, and examples for weather and climate tasks by matching GPU/VRAM requirements, forecast class, compatible data sources, and gallery examples.
Operational guide for enabling Megatron FSDP (Fully Sharded Data Parallel) in Megatron Bridge as an alternative to DDP, covering configuration knobs, sharding strategies, checkpoint format, and common pitfalls.
Covers resiliency features in Megatron Bridge including fault tolerance with in-job restarts, straggler detection, heartbeat-based hang detection, and automatic preemption-safe job resubmission via Slurm.
Guides users through downloading weather and climate data via Earth2Studio data source APIs, identifying compatible sources through the lexicon system and producing working xarray DataArray fetch scripts.
Stage 2 of NVIDIA's Clinical ASR Flywheel for curating clinical terms, tagging IPA pronunciations, and synthesizing NeMo manifests with TTS to build clinical speech recognition benchmarks.
An NVIDIA VSS Blueprint skill for searching video archives as part of the Video Search and Summarization platform, enabling AI-powered video content retrieval.
Guides enabling TP, DP, and PP communication overlap in Megatron Bridge to hide collective communication latency, including sequence-parallel overlap, distributed optimizer gradient and parameter gather overlap.
An NVIDIA Megatron-Core skill for code linting and formatting in large-scale distributed training projects using model parallelism, pipeline parallelism, and mixed precision.
An NVIDIA VSS Blueprint skill for generating video analysis reports as part of the Video Search and Summarization platform.
Provides representative MoE training playbooks by hardware platform (H100, B200, GB200, GB300), summarizing parallelism patterns, throughput bands, and common tuning stacks for each platform.
Deploys, operates, and debugs the RTVI-CV 2D detection/tracking microservice for NVIDIA's Video Search and Summarization blueprint, covering REST API operations, stream management, and health monitoring.
Stage 4 of NVIDIA's Clinical ASR Flywheel for fine-tuning Parakeet TDT v2 using stock NeMo SFT when priority KER exceeds 0.3, including offline cycle N+1 re-evaluation.
Stage 3 of NVIDIA's Clinical ASR Flywheel that scores NeMo manifests and produces five-section KER leaderboards with by-IPA-source diagnostics for evaluating clinical speech recognition accuracy.
Guides packed sequence configuration in Megatron Bridge, distinguishing offline packed SFT for LLMs from in-batch packing for VLMs, with context parallelism constraints and PackedSequenceSpecs setup.
Deploys any VSS (Video Search and Summarization) profile—base, search, lvs, warehouse, alerts, or edge—using a compose-centric workflow with env overrides, dry-run review, and GPU setup verification.
An NVIDIA VSS Blueprint skill for managing alerts and incidents in the Video Search and Summarization platform for automated video monitoring.
Deploys and operates the standalone RT-VLM dense captioning microservice for NVIDIA's Video Search and Summarization, covering file uploads, caption generation, stream management, and chat-completions via REST API.
An NVIDIA VSS Blueprint skill for deploying video embedding models as part of the Video Search and Summarization platform for video analytics and search.
An NVIDIA VSS Blueprint skill for querying video analytics data as part of the Video Search and Summarization platform for insights and reporting.
An NVIDIA VSS Blueprint skill for managing video input/output and storage configurations in the Video Search and Summarization platform.
An NVIDIA Megatron-Core skill for creating well-structured issues in the Megatron-LM repository for large-scale distributed training projects.
An NVIDIA VSS Blueprint skill for setting up behavior analytics in the Video Search and Summarization platform for monitoring and incident detection.
An NVIDIA VSS Blueprint skill for generating video calibration configurations as part of the Video Search and Summarization platform for deploying and managing video analytics.
An NVIDIA Megatron-Core skill for running distributed training jobs on Slurm clusters with support for model parallelism, pipeline parallelism, and mixed precision.
NVIDIA's official, verified catalog of portable agent skills that teach AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are installable via the `npx skills add nvidia/skills` CLI and target Claude Code, Codex, Cursor, Kiro, and other AI coding agents.
An NVIDIA Megatron-Core skill for splitting large pull requests into smaller, focused PRs in the Megatron-LM distributed training project.
An NVIDIA-verified medical AI skill built on MONAI for extracting metadata from DICOM medical imaging files, supporting clinical workflows for medical image analysis.
An NVIDIA Megatron-Core skill for testing large-scale distributed training code using model parallelism, pipeline parallelism, and mixed precision configurations.
An NVIDIA-verified skill for session memory management in NeMo-RL, the RLHF training framework built on Ray. Supports GRPO, DPO, and SFT for LLMs and VLMs with FSDP2 and Megatron-Core integration.
Installs the NVIDIA Holoscan SDK via the NGC Docker container, selecting the correct CUDA/architecture tag for the host GPU and validating with bundled Python and C++ examples.
Guides NVIDIA Holoscan SDK installation by inspecting host hardware, OS, and CUDA driver, then recommending the optimal install method — NGC container, Debian apt, pip, Conda, or source build — across Ubuntu, RHEL, IGX Orin, Jetson, and DGX platforms.
An NVIDIA-verified medical AI skill built on MONAI that performs preflight validation checks on DICOM series data before downstream medical AI processing.
An NVIDIA skill for installing and setting up the Holoscan SDK via Conda, part of the official NVIDIA-verified agent skills for sensor-to-AI application development.
Builds the NVIDIA Holoscan SDK from source using the in-tree ./run script within a Docker container, producing a local install tree consumable as a CMake dependency.
An NVIDIA skill for installing the Holoscan SDK via Python wheel packages, part of the official NVIDIA-verified agent skills.
An NVIDIA Medical AI skill for brain MR image generation built on MONAI, part of the official NVIDIA-verified agent skills.
An NVIDIA-verified skill for automated research workflows in NeMo-RL, the RLHF training framework built on Ray. Supports GRPO, DPO, and SFT training pipelines for LLMs and VLMs with FSDP2 and Megatron-Core.
An NVIDIA Medical AI skill for fine-tuning brain MR image generation models built on MONAI, part of the official NVIDIA-verified agent skills.
Converts a single CT DICOM series folder to a Hounsfield Unit NIfTI volume with affine evidence, designed for non-clinical research use with multi-frame DICOM explicitly excluded.
An NVIDIA skill for installing the Holoscan SDK via Debian packages, part of the official NVIDIA-verified agent skills supporting multiple installation methods.
An NVIDIA Medical AI skill for CT image segmentation built on MONAI, part of the official NVIDIA-verified agent skills for medical imaging workflows.
An NVIDIA Medical AI skill for MR image synthesis built on MONAI, part of the official NVIDIA-verified agent skills for medical imaging.
An NVIDIA Medical AI skill for fine-tuning VAE-based medical image generation models built on MONAI, part of the official NVIDIA-verified agent skills.
An NVIDIA Medical AI skill for CT image generation using flow-based models built on MONAI, part of the official NVIDIA-verified agent skills.
An NVIDIA Medical AI skill for evidence-oriented chest X-ray reasoning and evaluation built on MONAI, part of the official NVIDIA-verified agent skills.
An NVIDIA Medical AI skill for fine-tuning CT segmentation models built on MONAI, part of the official NVIDIA-verified agent skills.
An NVIDIA Medical AI skill for combined CT and MR image segmentation built on MONAI, part of the official NVIDIA-verified agent skills.
Enforces documentation conventions for the NeMo-RL project, covering docs/index.md maintenance and Google-style docstring formatting for Sphinx compatibility.
Provides operational guidance for running NeMo-RL experiments on Brev instances, covering storage rules for separating code from generated artifacts, managing large files on /ephemeral volumes, and handling secrets and caches.
Part of the official NVIDIA-verified agent skills catalog, providing portable instruction sets that teach AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly.
Part of the official NVIDIA-verified agent skills catalog teaching AI agents NVIDIA CUDA-X libraries, AI Blueprints, and platform tools.
A skill for FPGA flashing on NVIDIA Holoscan Sensor Bridge devkits, guiding agents through Lattice and VB1940 hardware flashing procedures.
A skill for automating QA test-plan execution on NVIDIA Holoscan Sensor Bridge devkits, guiding agents through structured hardware testing workflows.
A skill for running example applications on NVIDIA Holoscan Sensor Bridge devkits, enabling agents to execute and manage HSB demo workflows.
An official NVIDIA-verified skill that guides launching and managing TAO Toolkit training workflows on NVIDIA GPU infrastructure, covering job orchestration, monitoring, and pipeline execution.
Guides AI agents through running NVIDIA TAO Toolkit workflows on local Docker environments, enabling fine-tuning and optimization of pretrained vision AI models on local GPU infrastructure.
An official NVIDIA-verified skill that guides the setup and configuration of GPU host environments for NVIDIA TAO Toolkit, including driver installation, container runtime, and compute resource preparation.
An official NVIDIA-verified skill that guides fine-tuning Hugging Face models using NVIDIA TAO Toolkit, covering model conversion, training configuration, and optimization for NVIDIA GPU infrastructure.
Lists all available capabilities and supported model architectures in NVIDIA TAO Toolkit, helping AI agents discover which of the 100+ pretrained vision AI models and workflows are available.
An official NVIDIA-verified skill that guides porting Hugging Face models to NVIDIA TAO Toolkit format, covering model conversion, compatibility checks, and optimization for NVIDIA deployment pipelines.
Fine-tunes NVIDIA Cosmos Embed vision embedding models via TAO Toolkit, allowing customization of this foundation model with domain-specific data for embedding-based vision tasks.
Sets up and runs TAO Toolkit's inference service for deploying optimized vision AI models in production, enabling real-time inference at the edge or in the cloud.
Runs NVIDIA TAO Toolkit's AutoML pipeline to automatically search for optimal hyperparameters and model configurations, streamlining the training of vision AI models without manual tuning.
Trains image classification models with NVIDIA TAO Toolkit's low-code microservices, enabling fine-tuning of pretrained classifiers with custom datasets for production-ready edge or cloud inference.
Fine-tunes CLIP (Contrastive Language-Image Pretraining) models through NVIDIA TAO Toolkit, customizing this multimodal foundation model with your own data for vision-language applications.
Executes the AutoML-powered DEFT (Deep Efficient Fine-Tuning) pipeline in NVIDIA TAO Toolkit, combining automated hyperparameter search with efficient fine-tuning for optimized vision model training.
Fine-tunes NVIDIA Cosmos Reason models using TAO Toolkit's low-code microservices, enabling customization of this reasoning-capable vision foundation model with your own data for edge or cloud deployment.
Performs root cause analysis on Visual ChangeNet results within NVIDIA TAO Toolkit, diagnosing detection performance and identifying failure patterns in visual change detection workflows.
Deploys and runs NVIDIA TAO Toolkit workflows on the Brev cloud platform, providing a streamlined way to fine-tune and optimize pretrained vision AI models on cloud GPU instances.
Deploys NVIDIA TAO Toolkit workflows on Kubernetes clusters, enabling scalable distributed training and optimization of vision AI models across container-orchestrated GPU infrastructure.
Trains Masked Auto Encoder (MAE) self-supervised vision models through NVIDIA TAO Toolkit, learning rich visual representations from unlabeled image data for downstream fine-tuning tasks.
Generates video reasoning annotations using NVIDIA TAO Toolkit's data generation capabilities, creating labeled datasets for training vision AI models on temporal reasoning and video understanding tasks.
Analyzes performance gaps in Visual ChangeNet models within NVIDIA TAO Toolkit, evaluating change detection accuracy and identifying areas where the model needs improvement on specific data distributions.
Converts datasets between supported formats in NVIDIA TAO Toolkit, transforming annotation files and image data to match the input requirements of different model architectures and training pipelines.
An official NVIDIA-verified skill that guides training foundation stereo vision models using NVIDIA TAO Toolkit, covering dataset preparation, training configuration, and model evaluation workflows.
Trains Grounding DINO open-set object detection models with NVIDIA TAO Toolkit, enabling text-prompted detection where models locate objects described in natural language using custom training data.
Runs the DEFT AOI (Automated Optical Inspection) pipeline in NVIDIA TAO Toolkit, automating defect detection workflows for manufacturing quality control using fine-tuned vision AI models.
Trains CenterPose object pose estimation models using NVIDIA TAO Toolkit, leveraging pretrained vision AI models that can be fine-tuned with custom data and exported for production deployment.
Trains pose classification models using NVIDIA TAO Toolkit to categorize body poses in images and video, building on pretrained pose estimation models for fitness tracking, ergonomics, and safety applications.
Trains re-identification (ReID) models through NVIDIA TAO Toolkit, fine-tuning pretrained models to recognize and track individuals across different camera views for surveillance and retail analytics.
Trains action recognition models with NVIDIA TAO Toolkit to classify human activities in video, fine-tuning pretrained models with custom data for applications like surveillance, sports analytics, and gesture control.
Trains DINO self-supervised vision transformer models via NVIDIA TAO Toolkit, learning powerful visual features without labeled data that can be transferred to downstream detection and segmentation tasks.
Trains metric learning recognition models with NVIDIA TAO Toolkit, learning embedding spaces for similarity-based visual recognition tasks such as re-identification, face verification, and product matching.
Executes NVIDIA TAO Toolkit workflows on the NVIDIA AI Platform, providing managed infrastructure for fine-tuning and optimizing 100+ pretrained vision AI models at scale.
Generates image grounding data using NVIDIA TAO Toolkit, creating region-text pair annotations that link visual regions to language descriptions for training grounded vision-language models.
Runs NVIDIA TAO Toolkit workflows on Slurm-managed HPC clusters, enabling large-scale distributed training of vision AI models across multi-node GPU environments in research and enterprise settings.
Mines AOI (Automated Optical Inspection) images within NVIDIA TAO Toolkit's data pipeline, selecting and curating the most informative samples from inspection datasets for efficient model training.
Generates referring expression annotations with NVIDIA TAO Toolkit, producing natural language descriptions that identify specific objects in images for training vision-language understanding models.
Trains Mask Auto Label models through NVIDIA TAO Toolkit, automating segmentation mask generation for vision AI datasets using pretrained models and low-code workflows.
Trains Visual ChangeNet models using NVIDIA TAO Toolkit for detecting changes between image pairs, applicable to satellite imagery analysis, construction monitoring, and industrial inspection workflows.
Trains PointPillars 3D object detection models through NVIDIA TAO Toolkit, processing LiDAR point cloud data for autonomous driving and robotics with optimized models for edge deployment.
Trains Fast Foundation Stereo depth estimation models via NVIDIA TAO Toolkit, optimizing pretrained stereo vision models with custom data for real-time depth perception in edge deployments.
Trains Deformable DETR object detection models with NVIDIA TAO Toolkit, leveraging deformable attention mechanisms for efficient end-to-end detection that can be exported for production deployment.
Trains Mask2Former universal segmentation models using NVIDIA TAO Toolkit, supporting panoptic, instance, and semantic segmentation with a unified architecture fine-tuned on custom datasets.
Trains NVPanoptix3D panoptic segmentation models through NVIDIA TAO Toolkit, producing unified 3D scene understanding with both semantic and instance-level segmentation for autonomous systems.
Trains NVDINOv2 self-supervised vision models using NVIDIA TAO Toolkit, leveraging NVIDIA's optimized DINOv2 variant for feature extraction and transfer learning across vision tasks.
Trains Depth Anything V2 monocular depth estimation models using NVIDIA TAO Toolkit, fine-tuning this state-of-the-art depth model with custom data for robust depth prediction across diverse scenes.
A skill for training, evaluating, and deploying RT-DETR (Real-Time Detection Transformer) models for 2D object detection, with support for distillation, quantization, and AutoML-enabled training workflows.
Trains OCRNet semantic segmentation models with NVIDIA TAO Toolkit, fine-tuning pretrained models for pixel-level scene understanding with custom data and exporting optimized models for deployment.
Trains Mask Grounding DINO models via NVIDIA TAO Toolkit, combining open-set object detection with instance segmentation to detect and segment objects from text prompts using custom training data.
Routes Visual ChangeNet sample data within NVIDIA TAO Toolkit's change detection pipeline, directing training samples to appropriate processing stages for visual change analysis workflows.
Trains optical inspection models using NVIDIA TAO Toolkit for manufacturing defect detection, fine-tuning pretrained vision models to identify anomalies and quality issues in production environments.
Trains BEVFusion multi-sensor fusion models through NVIDIA TAO Toolkit, combining camera and LiDAR data for 3D object detection in autonomous driving and robotics applications.
Validates dataset formats within NVIDIA TAO Toolkit before training, checking annotation consistency, file structure, and data integrity to prevent errors during model training and optimization.
Runs NVIDIA TAO Toolkit workflows on the Lepton AI cloud platform, offering serverless GPU access for fine-tuning and optimizing pretrained vision AI models without infrastructure management overhead.
Analyzes gaps in VLM (Vision Language Model) BCQ evaluation results within NVIDIA TAO Toolkit, identifying performance weaknesses and quality benchmarks to guide targeted model improvements.
Trains OCDNet text detection models with NVIDIA TAO Toolkit, fine-tuning pretrained models to accurately detect and localize text regions in natural scene images for OCR pipelines.
Trains OneFormer universal image segmentation models with NVIDIA TAO Toolkit, using a single model architecture for panoptic, semantic, and instance segmentation guided by task-conditioned queries.
Trains SegFormer transformer-based semantic segmentation models using NVIDIA TAO Toolkit, combining efficient hierarchical encoders with lightweight decoders for fast and accurate segmentation on custom data.
Trains Sparse4D temporal 3D object detection models using NVIDIA TAO Toolkit, processing multi-camera video sequences for accurate 3D detection in autonomous driving scenarios.
Provides a streamlined single-step training workflow in NVIDIA TAO Toolkit, simplifying the process of training vision AI models by consolidating configuration, training, and export into one unified command.
Part of NVIDIA's official verified skills collection for AI agents, providing portable instruction sets that teach agents how to use NVIDIA software optimally, including CUDA-X libraries, AI Blueprints, and platform tools.
An official NVIDIA-verified agent skill that teaches AI agents how to optimally use NVIDIA software including CUDA-X libraries, AI Blueprints, and platform tools, installable via the skills CLI.
Official NVIDIA-verified skills that teach AI agents how to optimally use NVIDIA software including CUDA-X libraries, AI Blueprints, and platform tools. Installable via the skills CLI for Claude Code, Codex, Cursor, and other agents, with skills continuously added and mirrored daily from their respective product repositories.
Official NVIDIA-verified agent skills that teach AI agents how to use NVIDIA software optimally, including CUDA-X libraries, AI Blueprints, and platform tools. Skills are portable instruction sets maintained and synced from NVIDIA product repos.
An NVIDIA-verified skill that teaches AI agents to use NVIDIA software optimally, including CUDA-X libraries, AI Blueprints, and platform tools. Skills are portable instruction sets maintained across NVIDIA product repos and synced to a central catalog.
An official NVIDIA-verified agent skill from NVIDIA's skill catalog that teaches AI agents how to optimally use NVIDIA software, including CUDA-X libraries, AI Blueprints, and platform tools. Skills are maintained in their respective product repos and synced daily via an automated pipeline.
Part of NVIDIA's official verified skills collection, this skill teaches AI agents to use NVIDIA software optimally, covering CUDA-X libraries, AI Blueprints, and platform tools through portable instruction sets.
Part of the official NVIDIA Agent Skills catalog, providing portable instruction sets that teach AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly.
Part of the official NVIDIA Agent Skills catalog, providing portable instruction sets that teach AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly.
Part of the official NVIDIA Agent Skills catalog, providing portable instruction sets that teach AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly.
Official NVIDIA-verified agent skills catalog that teaches AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly, with skills maintained across product repos and mirrored via automated sync.
An official NVIDIA-verified skill for performance optimization, teaching AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly for GPU-accelerated computing.
An official NVIDIA-verified agent skill focused on PyTorch CUDA Graphs performance optimization. Part of the NVIDIA Agent Skills catalog covering CUDA-X libraries, AI Blueprints, and platform tools, with skills maintained in respective product repos and synced daily.
Official NVIDIA-verified agent skills that teach AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are maintained in product repos and synced to this catalog daily.
An official NVIDIA-verified agent skill from the NVIDIA skills catalog, which provides portable instruction sets that teach AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Installable via the skills CLI and compatible with Claude Code, Gemini CLI, Codex, Cursor, and other AI coding agents.
Official NVIDIA-verified skills for AI agents that teach correct usage of CUDA-X libraries, AI Blueprints, and platform tools, synced daily from NVIDIA product repos and installable via the skills CLI.
Official NVIDIA agent skills that teach AI agents to use CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are portable instruction sets maintained in NVIDIA product repos and synced to this catalog daily.
Official NVIDIA-verified skills that teach AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly, maintained in product repos and synced daily to this public catalog.
An official NVIDIA-verified debugging skill for AI agents that provides guidance on diagnosing issues with NVIDIA CUDA-X libraries and platform tools. Part of NVIDIA's portable skill catalog installable via the skills CLI.
Official NVIDIA-verified skills that teach AI agents how to use CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are maintained in their respective product repos and synced to this catalog daily.
Official NVIDIA-verified skills that teach AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are maintained in product repos and mirrored daily via an automated sync pipeline.
Official NVIDIA-verified skills that teach AI agents how to use CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are portable instruction sets maintained in their respective product repos and mirrored daily via an automated sync pipeline.
An official NVIDIA-verified agent skill from the NVIDIA skills catalog, teaching AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Maintained in NVIDIA product repos and synced daily via automated pipeline.
An NVIDIA-verified skill that teaches AI agents to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools, installable via the skills CLI and compatible with multiple agent runtimes.
Official NVIDIA-verified skills that teach AI agents how to use CUDA-X libraries, AI Blueprints, and platform tools correctly, maintained in their respective product repos and synced daily.
An official NVIDIA-verified agent skill from the NVIDIA skills catalog, teaching AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Maintained in NVIDIA product repos and synced daily via automated pipeline.
Part of NVIDIA's official, verified skills for AI agents, providing portable instruction sets that teach agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly.
An official NVIDIA-verified skill that teaches AI agents to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly, installable via the skills CLI and compatible with Claude Code, Cursor, Codex, and other AI coding agents.
An official NVIDIA-verified skill for AI agents, teaching them to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Installable through the standard skills CLI from the NVIDIA skills catalog.
An official NVIDIA-verified skill from the NVIDIA Agent Skills catalog that teaches AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools. Skills are maintained in their respective product repos and synced here via an automated pipeline.
Official NVIDIA-verified skills that teach AI agents how to use CUDA-X libraries, AI Blueprints, and platform tools correctly, distributed as portable instruction sets following the Agent Skills specification.
Official NVIDIA-verified agent skills that teach AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly, installable via the skills CLI.
Part of the NVIDIA Agent Skills catalog, this skill guides AI agents through launching LLM evaluations using the NeMo Evaluator Launcher, accessing MLflow results, and running bring-your-own benchmarks.
An NVIDIA-verified skill from the Video Search and Summarization (VSS) Blueprint that teaches AI agents to deploy profiles, search and summarize video content, generate analysis reports, manage alerts and incidents, and query VIOS sensors.
An NVIDIA-verified skill that teaches AI agents how to access and retrieve MLflow experiment results within the NeMo Evaluator LLM evaluation pipeline.
An official NVIDIA-verified skill for AI agents that teaches how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools. Skills are synced daily from NVIDIA product repos and installable via the skills CLI.
Official NVIDIA-verified agent skills that teach AI agents to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are maintained in their respective product repos and mirrored daily via an automated sync pipeline, installable through the standard skills CLI.
An NVIDIA-verified skill from the TileGym project that guides AI agents through tile-based GPU programming, including adding new kernels, cross-framework conversion, and performance optimization.
An NVIDIA-verified skill from the official NVIDIA agent skills catalog that guides AI agents through tile-based GPU kernel optimization using the TileGym/TileIR framework, covering kernel authoring, cross-framework conversion, and performance tuning.
An official NVIDIA skill that teaches AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools through portable instruction sets covering optimization, numerical computing, and routing formulation.
Official NVIDIA-verified agent skills that teach AI agents how to use CUDA-X libraries, AI Blueprints, and platform tools correctly, with skills maintained in product repos and synced daily to this catalog.
An NVIDIA skill from the Video Search and Summarization (VSS) Blueprint that guides AI agents through video content summarization workflows, including deploying profiles and generating analysis reports.
An official NVIDIA-verified skill that teaches AI agents how to explore and work with TensorRT-LLM codebases, part of the NVIDIA skills catalog for CUDA-X libraries and AI platform tools.
Part of the official NVIDIA Agent Skills catalog, providing portable instruction sets that teach AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are maintained in product repos and mirrored daily via an automated sync pipeline.
An official NVIDIA-verified skill that teaches AI agents CPU offloading performance optimization using NVIDIA CUDA-X libraries, part of the NVIDIA Agent Skills catalog synced from product repos.
An NVIDIA skill from the Video Search and Summarization (VSS) Blueprint that provides guidance for using the RTVI VLM (Vision Language Model) microservice for real-time video understanding and analysis.
An official NVIDIA-verified agent skill from the NVIDIA skills catalog. Provides guidance on host-side performance optimization for NVIDIA CUDA-X libraries and platform tools, installable via the standard skills CLI.
Part of NVIDIA's official agent skills for the Video Search and Summarization Blueprint, supporting video search, summarization, analysis report generation, alert and incident management, and VIOS sensor queries.
An official NVIDIA-verified skill that teaches AI agents how to create issues using NVIDIA platform tools. Part of the NVIDIA Agent Skills catalog covering CUDA-X libraries and AI Blueprints.
Part of the official NVIDIA Agent Skills catalog of 155+ verified skills covering CUDA-X libraries, AI Blueprints, and platform tools, maintained across product repos including TensorRT-LLM, Megatron-Core, and other NVIDIA technologies.
Official NVIDIA-verified skills for AI agents, providing portable instruction sets that teach agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are installable via the standard skills CLI and are mirrored daily from their respective NVIDIA product repos.
Part of the official NVIDIA-verified skills catalog that teaches AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are mirrored from product repos via an automated daily sync pipeline.
Part of the official NVIDIA Agent Skills catalog, providing portable instruction sets that teach AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are mirrored daily from product repos and installable via the skills CLI.
Official NVIDIA-verified skills that teach AI agents how to use CUDA-X libraries, AI Blueprints, and NVIDIA platform tools correctly, installable via the skills CLI and compatible with Claude Code, Cursor, and other agents.
Official NVIDIA-verified agent skills that teach AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly, installable via the skills CLI for Claude Code, Cursor, Codex, and other AI coding agents.
An NVIDIA-verified skill that teaches AI agents performance sequence packing techniques for CUDA-X libraries. Part of the official NVIDIA Agent Skills catalog with portable instruction sets for AI Blueprints and platform tools.
Part of NVIDIA's official agent skills catalog, providing portable instruction sets that teach AI agents to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are maintained in their respective product repos and mirrored daily via an automated sync pipeline.
Official NVIDIA-verified skills catalog that teaches AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are maintained in their respective product repositories and mirrored daily via an automated sync pipeline.
An official NVIDIA-verified skill that teaches AI agents to use activation recomputation techniques for performance optimization in CUDA-X workflows. Part of the NVIDIA skills catalog covering CUDA-X libraries, AI Blueprints, and platform tools.
An NVIDIA NeMo Evaluator skill that lets users bring their own custom benchmarks for LLM evaluation, alongside built-in evaluation workflows and MLflow result access.
An official NVIDIA-verified skill from the NVIDIA Agent Skills catalog that teaches AI agents how to correctly use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools through portable instruction sets.
Part of the official NVIDIA Agent Skills catalog, providing portable instruction sets that teach AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are maintained in NVIDIA product repos and mirrored daily via automated sync.
Part of NVIDIA's official, verified agent skills catalog that teaches AI agents how to correctly use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools through portable instruction sets.
An official NVIDIA-verified skill that teaches AI agents how to correctly use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools for generating reports and analysis.
An official NVIDIA-verified skill from the NVIDIA Agent Skills catalog that provides portable instruction sets for AI agents to correctly compile and execute code using NVIDIA CUDA-X libraries and platform tools.
Part of the official NVIDIA Agent Skills catalog, providing portable instruction sets that teach AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly.
Part of NVIDIA's official verified skills catalog that teaches AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools. Skills are maintained in product repos and synced daily, installable via the skills CLI for Claude Code, Cursor, Codex, and other agents.
Part of the official NVIDIA Agent Skills catalog, providing portable instruction sets that teach AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly.
A skill from NVIDIA's official catalog that teaches AI agents to use CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are portable instruction sets mirrored daily from NVIDIA product repos and installable via the skills CLI.
Part of the official NVIDIA Agent Skills catalog, providing portable instruction sets that teach AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly.
Part of the official NVIDIA Agent Skills catalog, providing portable instruction sets that teach AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are maintained in NVIDIA product repos and mirrored daily via automated sync.
Official NVIDIA-verified skills that teach AI agents how to correctly use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools, installable via the skills CLI and compatible with Claude Code, Cursor, and other agents.
Part of the official NVIDIA Agent Skills catalog, providing portable instruction sets that teach AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly.
Part of NVIDIA's official verified skills catalog that teaches AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools. Skills are maintained in product repos and synced daily, installable via the skills CLI for Claude Code, Cursor, Codex, and other agents.
An official NVIDIA-verified skill from the NVIDIA Agent Skills catalog that provides portable instruction sets for AI agents to work with NVIDIA NeMo and related platform tools for maintainer workflows.
Part of NVIDIA's official verified skills catalog that teaches AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools. Skills are maintained in product repos and synced daily, installable via the skills CLI for Claude Code, Cursor, Codex, and other agents.
An official NVIDIA-verified agent skill that provides AI coding assistants with instructions for using NVIDIA CUDA-X libraries, AI Blueprints, and platform tools, installable via the Skills CLI for Claude Code, Cursor, GitHub Copilot, and other agents.
Official NVIDIA-verified agent skills that teach AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are mirrored daily from NVIDIA product repos and installable via the skills CLI for Claude Code, Cursor, Codex, and Kiro.
Part of NVIDIA's official verified agent skills catalog, providing portable instruction sets that teach AI agents to correctly use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools.
Part of NVIDIA's official agent skills catalog, a curated set of portable instruction sets maintained across NVIDIA product repos and mirrored daily, teaching AI agents to work with CUDA-X libraries and AI Blueprints.
Part of NVIDIA's official verified skills catalog that provides portable instruction sets teaching AI agents to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools. Installable via the skills CLI and compatible with Claude Code, Cursor, and other AI coding agents.
An NVIDIA-verified agent skill from the official NVIDIA skills catalog, providing portable instruction sets that teach AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools.
Official NVIDIA-verified skills that teach AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are maintained in product repos and mirrored daily via an automated sync pipeline.
Part of NVIDIA's official verified skills catalog that provides portable instruction sets teaching AI agents to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools. Installable via the skills CLI and compatible with Claude Code, Cursor, and other AI coding agents.
Part of NVIDIA's official agent skills catalog providing portable instruction sets for AI agents to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are maintained in NVIDIA product repos and mirrored daily via an automated sync pipeline.
Official NVIDIA-verified skills for AI agents that teach how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools, installable via the skills CLI with support for Claude Code, Cursor, Codex CLI, Gemini CLI, and other AI coding assistants.
Part of NVIDIA's official verified skills catalog that provides portable instruction sets teaching AI agents to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools. Installable via the skills CLI and compatible with Claude Code, Cursor, and other AI coding agents.
Official NVIDIA-verified agent skills that teach AI agents to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are mirrored from NVIDIA product repos and cover areas like numerical optimization (cuOpt), performance tuning, and MoE dispatcher selection.
Part of NVIDIA's official agent skills catalog providing portable instruction sets for AI agents to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are maintained in NVIDIA product repos and mirrored daily via an automated sync pipeline.
Part of the official NVIDIA Agent Skills catalog, which provides NVIDIA-verified instruction sets that teach AI agents how to use CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are mirrored daily from their respective product repos via an automated sync pipeline.
Part of NVIDIA's official verified skills catalog that provides portable instruction sets teaching AI agents to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools. Installable via the skills CLI and compatible with Claude Code, Cursor, and other AI coding agents.
Part of NVIDIA's NemoClaw skill collection for secure agent sandboxing, enabling OpenClaw to run inside NVIDIA OpenShell with managed inference, policy management, remote deployment, sandbox monitoring, and contributor/maintainer workflows.
An official NVIDIA-verified agent skill from the NVIDIA skills catalog for managing dependency updates. Part of a collection of portable instruction sets that teach AI agents to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly, installable via the skills CLI.
An NVIDIA-verified agent skill that provides portable instructions for AI agents to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Part of the official NVIDIA skills catalog synced daily from product repos.
Official NVIDIA-verified skills that teach AI agents to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are maintained in their respective product repos and mirrored daily, installable via the skills CLI.
Official NVIDIA-verified agent skills that teach AI coding agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools. Skills are maintained in their respective NVIDIA product repos and mirrored daily via an automated sync pipeline.
Part of NVIDIA Agent Skills, an official catalog of NVIDIA-verified skills that teach AI agents how to use CUDA-X libraries, AI Blueprints, and platform tools. Skills are maintained in their respective product repos and mirrored daily via an automated sync pipeline.
Part of NVIDIA's official, verified agent skills catalog that teaches AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools. Skills are maintained in their respective product repos and mirrored daily via an automated sync pipeline.
Part of the official NVIDIA Agent Skills catalog, providing portable instruction sets that teach AI agents how to correctly use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools.
An official NVIDIA-verified agent skill from the NVIDIA skills catalog for add-fusion graph transformations. Part of a collection of portable instruction sets that teach AI agents to use NVIDIA CUDA-X libraries and deep learning compiler optimizations, installable via the skills CLI.
An NVIDIA-verified agent skill that provides portable instructions for AI agents to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Part of the official NVIDIA skills catalog maintained across product repos.
An NVIDIA-verified agent skill that provides portable instructions for optimizing MoE (Mixture of Experts) communication overlap performance using NVIDIA CUDA-X libraries. Installable via the skills CLI for use with Claude Code, Cursor, Codex, and other compatible agents.
An official NVIDIA-verified skill from the NVIDIA Agent Skills catalog that provides AI agents with guidance on NVIDIA CUDA-X libraries and platform tools. Skills are maintained in their respective product repos and mirrored daily via an automated sync pipeline.
Provides practical guidance for training Mixture-of-Experts vision-language models in Megatron Bridge, comparing FSDP and 3D-parallel approaches with lessons from Qwen3-VL and other multimodal experiments.
An official NVIDIA-verified skill from the NVIDIA Agent Skills catalog that provides AI agents with guidance on NVIDIA CUDA-X libraries and platform tools. Skills are maintained in their respective product repos and mirrored daily via an automated sync pipeline.
An NVIDIA-verified agent skill that provides portable instructions for responding to GitHub issues using NVIDIA tools and workflows. Installable via the skills CLI for use with Claude Code, Cursor, Codex, and other compatible agents.
Part of the official NVIDIA Agent Skills catalog that teaches AI agents to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools, covering areas like large-scale distributed training with Megatron-Core and FSDP.
An NVIDIA-verified agent skill from the official catalog of portable instruction sets that teach AI agents to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly.
An official NVIDIA-verified agent skill from the NVIDIA Skills catalog. The catalog provides portable instruction sets that teach AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly, with skills maintained in their respective product repos and synced daily.
An official NVIDIA-verified agent skill that provides AI coding assistants with instructions for using NVIDIA CUDA-X libraries, AI Blueprints, and platform tools, installable via the Skills CLI for Claude Code, Cursor, GitHub Copilot, and other agents.
Part of NVIDIA's official verified skills repository for AI agents, providing portable instruction sets that teach agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly, with skills maintained in product repos and mirrored daily.
Guides running Megatron-LM and Megatron Bridge training with mock or real data, covering correlation testing between MLM and Bridge loss curves, available recipes, and multi-GPU examples.
An NVIDIA-verified agent skill for debugging NeMo Gym reinforcement learning training environments. Part of a catalog of portable skills covering CUDA-X libraries, AI Blueprints, and platform tools, synced daily from NVIDIA product repos.
An official NVIDIA-verified skill for reward profiling in NeMo Gym RL training environments, part of the NVIDIA skills catalog that teaches AI agents to use CUDA-X libraries and platform tools correctly.
Official, NVIDIA-verified skills that teach AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are maintained in their respective product repos and mirrored here daily via an automated sync pipeline.
A suite of 113 AI agent skills for quantitative trading, market intelligence, and creative production, optimized for Claude and Gemini agents.
Official NVIDIA-verified skills for AI agents, providing portable instruction sets that teach agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly.
Official NVIDIA-verified skills that teach AI agents how to correctly use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools, maintained in their respective product repos and mirrored daily through an automated sync pipeline.
Official NVIDIA-verified skill that teaches AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are maintained in NVIDIA product repos and synced to this catalog daily via an automated pipeline.
Part of the official NVIDIA Agent Skills repository, providing NVIDIA-verified portable instruction sets that teach AI agents to correctly use CUDA-X libraries, AI Blueprints, and platform tools. Skills are mirrored daily from product repos and installable via the skills CLI for Claude Code, Codex, Cursor, and Kiro.
Part of the official NVIDIA-verified skills collection that teaches AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools. Skills are maintained in respective product repos and mirrored daily via automated sync.
Official NVIDIA-verified agent skills for working with CUDA-X libraries, AI Blueprints, and platform tools. Skills are maintained in their respective product repos and mirrored to this catalog.
Official, NVIDIA-verified skills for AI agents that teach how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly. Skills are portable instruction sets mirrored daily from product repos, installable via the skills CLI for Claude Code, Codex, Cursor, and other agents.
An official NVIDIA-verified agent skill from the NVIDIA Agent Skills collection. Built on the Agent Skills specification and designed for AI coding agents across multiple platforms.
Skill
Skill
Skill
Official NVIDIA-verified agent skills for AI coding assistants, following the Agent Skills specification standard. Provides domain-specific capabilities with Apache 2.0 licensing.
Skill
Official NVIDIA-verified skills that teach AI agents how to use NVIDIA CUDA-X libraries, AI Blueprints, and platform tools correctly, installable via the skills CLI with support for Claude Code, Cursor, and other agents.
Skill
Skill
Skill
Skill
Skill
Skill
Skill
Skill
Skill
Skill
Skill
Skill
Skill
Skill
Skill
Skill
Skill
Skill
Skill
Skill
Skill
Skill
Skill
Skill
Skill
Skill
Skill
Skill
Skill