NeVibeCode · repository · 312 items
skills: a set by nvidia
The nvidia/skills repository on GitHub: 312 cards with a description and an install command. Inside: 311 skills and 1 plugin. They install into Claude Code, Codex and Cursor.
About the repository
- GitHub
- nvidia/skills
- Stars
- ★ 4,106
- Installs across cards
- 334,469
- Licenses
- Apache-2.0, CC-BY-4.0, LicenseRef-Proprietary
- Source updated
- 2026-11-20
- In the catalog since
- 2026-10-25
rag-blueprint skill This skill deploys, configures, troubleshoots, and manages NVIDIA RAG Blueprint. Use it when RAG services run through Docker, Kubernetes, or library workflows.
CX Codex +1
3,500installs
accelerated-computing-cudf skill accelerated-computing-cudf is a Claude Code skill that guides cuDF and dask-cuDF use to move pandas-style data work to GPUs and check result semantics.
CX Codex +1
3,200installs
cuopt-developer skill cuopt-developer guides AI coding assistants through modifying and building NVIDIA cuOpt internals, covering C++/CUDA and Python bindings.
CX Codex +1
2,800installs
dynamo-recipe-runner skill dynamo-recipe-runner is a Claude Code skill that finds, validates, patches, and deploys NVIDIA Dynamo Kubernetes recipes for model serving with vLLM or SGLang.
CX Codex +1
2,600installs
cuopt-install skill cuopt-install installs NVIDIA cuOpt for Python, C, or as a REST server via pip, conda, or Docker, then verifies the setup. Helpful when configuring GPU-accelerated routing environments.
CX Codex +1
2,600installs
data-designer skill The data-designer skill builds synthetic datasets and Python generation pipelines from a plain-language brief. It starts when an agent needs to turn a dataset description into a runnable configuration.
CX Codex +1
2,600installs
deepstream-dev skill deepstream-dev teaches the agent to write NVIDIA DeepStream SDK code in Python: GStreamer pipelines, TensorRT inference, tracking and broker output.
CX Codex +1
2,500installs
nemo-retriever skill Indexes and queries document collections using the NeMo Retriever 26.8.1 CLI. Useful for local LanceDB storage and deployed Retriever services.
CX Codex +1
2,500installs
rag-eval skill rag-eval evaluates filesystem RAG datasets built around corpus/ and train.json. Use it to compare retrieval and generation settings with RAGAS quality scores.
CX Codex +1
2,500installs
cudaq-guide skill cudaq-guide walks you through CUDA-Q installation, simulation targets, QPU access and Python kernel authoring. Useful when you need to get moving with NVIDIA's quantum stack.
CX Codex +1
2,500installs
omniverse-usd-performance-tuning skill omniverse-usd-performance-tuning is a Claude Code skill that diagnoses and optimizes USD stages with slow loading, high memory use, low FPS or validation failures.
CX Codex +1
2,400installs
omniverse-realtime-viewer skill The omniverse-realtime-viewer Claude Code skill routes work on Omniverse USD viewer apps to the relevant architecture and implementation references. It covers XR delivery, GPU processing, and stylized rendering paths.
CX Codex +1
2,300installs
skill-card-generator skill skill-card-generator turns a skill folder that already exists into a draft governance card: it reads the files, writes a context, renders markdown and checks that review markers are gone.
CX Codex +1
2,300installs
cuopt-routing-api-python skill cuopt-routing-api-python — a skill that walks your agent through vehicle routing with the cuOpt Python API: TSP, VRP and pickup-delivery. Useful when the solver runs in Python.
CX Codex +1
2,300installs
deepstream-import-vision-model skill The deepstream-import-vision-model skill takes an object-detection model from HuggingFace or NGC through a full NVIDIA DeepStream pipeline: ONNX, TRT engine, bbox parser, benchmark, PDF report.
CX Codex +1
2,300installs
cuopt-numerical-optimization-formulation skill cuopt-numerical-optimization-formulation turns a written problem into a clean LP, MILP or QP formulation: variables, constraints, objective. Concepts only, no API code.
CX Codex +1
2,200installs
nemotron-customize skill nemotron-customize is a Claude Code skill that plans Nemotron customization workflows. It activates for requests to train, evaluate, convert, or otherwise configure Nemotron models.
CX Codex +1
2,200installs
physical-ai-neural-reconstruction skill The physical-ai-neural-reconstruction skill routes NVIDIA NuRec and NRE requests to the right companion skill for datasets, conversion, training, or rendering. Use it to choose the next step in a 3D reconstruction workflow.
CX Codex +1
2,200installs
cuopt-server-api-python skill cuopt-server-api-python deploys the NVIDIA cuOpt REST server and guides client requests via Python or curl. Useful for solving routing, LP, and MILP optimization problems on GPUs.
CX Codex +1
2,200installs
rag-perf skill rag-perf runs a profiling pass and an aiperf load test against a deployed NVIDIA RAG Blueprint server, all driven by one YAML config.
CX Codex +1
2,200installs
nemotron-retrieval-recipes skill nemotron-retrieval-recipes helps plan, debug, tune, export, and deploy public NVIDIA Nemotron embedding and reranking retrieval recipes. Useful for fixing recall and ranking issues in search pipelines.
CX Codex +1
2,100installs
nemo-rl-session-memory skill nemo-rl-session-memory keeps a durable, human-readable record of a working session under session/, so another agent can resume after a disconnect or a VS Code restart.
CX Codex +1
2,100installs
nemotron-speech skill Walks your agent through NVIDIA Nemotron Speech (formerly Riva) NIM work: deploying, running and testing ASR, TTS and NMT in the cloud or self-hosted.
CX Codex +1
2,100installs
dali-dynamic-mode skill dali-dynamic-mode — a skill that steers the agent through NVIDIA DALI's imperative mode: direct ndd calls instead of pipeline graphs, plus help migrating existing pipelines.
CX Codex +1
2,100installs
nemo-automodel-model-onboarding skill nemo-automodel-model-onboarding walks you through adding a new model architecture to NeMo AutoModel, from reading config.json to registering the class and writing tests.
CX Codex +1
2,000installs
nemo-automodel-recipe-development skill nemo-automodel-recipe-development guides the agent in creating and editing NeMo AutoModel training and evaluation recipes. It covers YAML configuration, builder functions, and CLI execution flow.
CX Codex +1
2,000installs
nemo-automodel-distributed-training skill nemo-automodel-distributed-training picks a distributed training strategy for NeMo AutoModel and lists the YAML fields to set. Useful when the model no longer fits on one GPU.
CX Codex +1
2,000installs
physical-ai-defect-image-generation skill physical-ai-defect-image-generation prepares automated defect synthesis pipelines for optical quality inspection. Helps construct training sets for electronic boards, metal, and glass using Cosmos vision models.
CX Codex +1
2,000installs
physical-ai-video-data-augmentation skill physical-ai-video-data-augmentation is a Claude Code skill for OSMO video augmentation and auto-labeling jobs. It selects a workflow, checks readiness, tracks execution, and retrieves outputs.
CX Codex +1
2,000installs
tilegym-adding-cutile-kernel skill tilegym-adding-cutile-kernel guides AI coding agents through adding cuTile GPU kernel operators to TileGym. Useful when implementing custom kernels, tests, and benchmarks.
CX Codex +1
1,900installs
nemo-mbridge-perf-moe-optimization-workflow skill nemo-mbridge-perf-moe-optimization-workflow guides MoE throughput tuning and regression diagnosis in Megatron Bridge. Useful when diagnosing speed drops after commit or configuration changes.
CX Codex +1
1,900installs
nemo-mbridge-multi-node-slurm skill nemo-mbridge-multi-node-slurm is a Claude Code skill that turns single-node scripts into multi-node Slurm jobs and helps diagnose launch and NCCL failures.
CX Codex +1
1,900installs
nemo-mbridge-perf-moe-vlm-training skill nemo-mbridge-perf-moe-vlm-training is a Claude Code skill with practical guidance for training MoE VLMs in Megatron Bridge, configuring FSDP or 3D parallelism, and debugging OOM failures.
CX Codex +1
1,900installs
nemo-mbridge-mlm-bridge-training skill The nemo-mbridge-mlm-bridge-training skill runs Megatron-LM and Megatron Bridge training, compares loss curves, and helps investigate divergence after code changes.
CX Codex +1
1,900installs
nemo-mbridge-perf-parallelism-strategies skill nemo-mbridge-perf-parallelism-strategies is a Claude Code skill for selecting parallelism settings in Megatron Bridge. It activates for questions about TP, DP, PP, CP, or EP, and when tracing OOMs or regressions to config changes.
CX Codex +1
1,900installs
nemo-mbridge-perf-activation-recompute skill nemo-mbridge-perf-activation-recompute guides activation recomputation setup in Megatron Bridge to resolve GPU memory OOMs during LLM training. Balances memory savings against extra recomputed operations.
CX Codex +1
1,900installs
dynamo-troubleshoot skill dynamo-troubleshoot is a Claude Code skill that investigates unhealthy Dynamo deployments. It activates for failures in pods, PVCs, model-cache jobs, workers, endpoints, or benchmarks.
CX Codex +1
1,900installs
nemo-mbridge-recipe-recommender skill nemo-mbridge-recipe-recommender — a skill for Claude Code selects and tunes Megatron Bridge training recipes based on model architecture, GPU count, and workload. Configures TP, CP, and SP layout for pretraining, SFT, or PEFT.
CX Codex +1
1,900installs
launch-nemo-rl skill launch-nemo-rl is a Claude Code skill for launching, monitoring, and debugging NeMo-RL recipes on Kubernetes with the nrl-k8s CLI.
CX Codex +1
1,900installs
nemo-mbridge-perf-memory-tuning skill nemo-mbridge-perf-memory-tuning is a Claude Code skill for lowering GPU memory peaks and diagnosing OOM during Megatron Bridge training. It covers allocator settings, parallelism, and activation memory.
CX Codex +1
1,900installs
1–40 of 312