SkillAtlasSkill 详情

doca-dpa

Official, NVIDIA-verified Agent Skills for Claude Code, Codex, and other coding agents.

审核状态:已审核Quality 72Security 70

复制安装命令

用 Codex 或 Claude 安装复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它先审查 Skill 页面再帮你安装。

复制前请先查看来源、License 和安全提示。

项目 README

来源文件:README.md

抓取于 2026年8月9日

NVIDIA Agent Skills

Official, NVIDIA-verified Agent Skills for Claude Code, Codex, and other coding agents.

NVIDIA Agent Skills Spec License

📖 Docs: docs.nvidia.com/skills  ·  📺 Livestream: From Vulnerable to Verified  ·  📝 Blog: NVIDIA Verified Agent Skills: Capability Governance for AI Agents


Skills are portable instruction sets that teach AI agents how to use NVIDIA software optimally: Physical AI and robotics workflows, simulation, CUDA-X libraries, RAG and AI Blueprints, and platform tools. This repository is a catalog: skills are maintained in their respective product repos, and mirrored here daily via an automated sync pipeline. Skills are being added continuously, so check back for updates. We are building this infrastructure in the open, and contributions are welcome. See the Roadmap for what is planned next.


Quickstart

Install NVIDIA skills with the default skills CLI flow:

npx skills add nvidia/skills

The CLI runs through npx and prompts you to choose a skill and install destination. You do not need to clone this repo or copy skill folders by hand.

Requires a current skills CLI (v1.5.16 or newer). Installing via npx skills@latest add nvidia/skills always uses the latest. On older CLIs (v1.5.15 and earlier), skills may install but not appear in Claude Code — see Troubleshooting.

The skill is available the next time your agent loads skills and encounters a relevant task. For example, ask your agent to "solve a linear programming problem with cuOpt" and the skill guides it through the cuOpt Python API. In Claude Code, run /reload-skills to load newly installed skills in your current session.

Install One Skill Without Prompts

Use this when you already know the skill name and want to skip prompts.

npx skills add nvidia/skills --skill cuopt-numerical-optimization-api --yes

Replace cuopt-numerical-optimization-api with any skill name from the Skill Catalog.

Install for a Specific Agent

Use --agent to target a specific AI coding agent. Initially, we'll support common client targets, expanding the list over time. For the full list of clients supported by the spec, see the skills CLI Supported Agents table.

Claude Code

npx skills add nvidia/skills --skill cuopt-numerical-optimization-api --agent claude-code

Codex

npx skills add nvidia/skills --skill cuopt-numerical-optimization-api --agent codex

Snowflake CoCo

npx skills add nvidia/skills --skill cuopt-numerical-optimization-api --agent cortex

Cursor

npx skills add nvidia/skills --skill cuopt-numerical-optimization-api --agent cursor

Kiro

npx skills add nvidia/skills --skill cuopt-numerical-optimization-api --agent kiro-cli

Use --agent more than once to install the same skill into multiple agents.

npx skills add nvidia/skills \
  --skill cuopt-numerical-optimization-api \
  --agent claude-code \
  --agent codex \
  --agent cursor \
  --agent kiro-cli

Keep Skills Up to Date

New skills land continuously, and existing ones are revised, renamed, or consolidated as the catalog evolves. Refresh what you have installed with:

npx skills update

Run it interactively and the CLI also flags skills that were removed or merged upstream (for example, when several skills are consolidated into one) and offers to remove the stale local copies. Use npx skills list to see what is installed and npx skills check to preview what is out of date first.

Browse the Catalog

Use this when you want to see available NVIDIA skills before installing anything.

npx skills add nvidia/skills --list

For non-interactive installs, global installs, agent-specific installs, updates, removals, and fallback manual copying, see Advanced installation.


Skill Catalog

ProductDescriptionSkills
AIQNVIDIA AI-Q Blueprint - deploy local AI-Q services and run shallow or deep research workflows as agent skills.aiq-research, aiq-deploy
CUDA-QCUDA Quantum — onboarding guide for installation, test programs, GPU simulation, QPU hardware, and quantum applications.cudaq-guide
cuDFOfficial NVIDIA-authored guidance for NVIDIA cuDF GPU DataFrames, pandas acceleration, dask-cuDF, ETL, joins, groupby, CSV/Parquet I/O, nullable semantics, and multi-GPU DataFrame workloads.accelerated-computing-cudf
cuOptGPU-accelerated optimization — vehicle routing, linear programming, quadratic programming, installation, server deployment, and developer tools.cuopt-install, cuopt-multi-objective-exploration, cuopt-numerical-optimization-api, cuopt-numerical-optimization-formulation, cuopt-routing-api-python, cuopt-server-api-python
cuPyNumericNumPy and SciPy on multi-node multi-GPU systems — skills to help with installing cuPyNumeric, migrating existing NumPy code, and doing parallel I/Ocupynumeric-hdf5, cupynumeric-install, cupynumeric-migration-readiness, cupynumeric-parallel-data-load
DALIGPU-accelerated data loading and processing with NVIDIA DALI.dali-dynamic-mode
Data DesignerBuild declarative synthetic dataset generation pipelines with NeMo Data Designer.data-designer
DeepStreamAgentic skills for guided DeepStream development.amc-run-sample-calibration, amc-run-video-calibration, amc-setup-calibration-stack, deepstream-dev, deepstream-generate-pipeline, deepstream-import-vision-model, deepstream-profile-pipeline, deepstream-sop
Digital HealthAgent skills for the clinical ASR evaluation flywheel — term curation, synthetic clinical-speech benchmark generation, KER (Keyword Error Rate) scoring, and fine-tune guidance.digital-health-clinical-asr-setup, digital-health-clinical-asr-build, digital-health-clinical-asr-eval, digital-health-clinical-asr-finetune
DOCATeach AI agents to use the NVIDIA DOCA SDK on BlueField DPUs and ConnectX NICs — setup, libraries, services, tools, deployment, and debugging.doca-bare-metal-deployment, doca-bf3-deployment, doca-bf4-deployment, doca-collectx-deployment, doca-container-deployment, doca-debug, doca-hardware-safety, doca-programming-guide, doca-public-knowledge-map, doca-setup, doca-structured-tools-contract, doca-upgrade, doca-version, doca-aes-gcm, doca-argp, doca-comch, doca-common, doca-compress, doca-devemu, doca-dma, doca-dpa, doca-dpdk-bridge, doca-erasure-coding, doca-eth, doca-flow, doca-flow-dpa-provider, doca-gpi, doca-gpunetio, doca-mgmt, doca-pcc, doca-pcc-ztr-rttcc-algo, doca-rdma, doca-rdmi, doca-rmax, doca-sha, doca-sta, doca-telemetry, doca-telemetry-exporter, doca-urom, doca-verbs, doca-argus, doca-dms, doca-firefly, doca-urom-svc, doca-bench, doca-bench-extension, doca-caps, doca-comm-channel-admin, doca-dpa-hl-tracer, doca-flow-dpa-perf, doca-flow-grpc-server, doca-flow-perf, doca-flow-tune, doca-gpunetio-ib-write-bw, doca-gpunetio-ib-write-lat, doca-pcc-counters, doca-sha-offload-engine, doca-socket-relay, doca-spcx-cc, doca-telemetry-utils
DynamoNVIDIA Dynamo deployment bring-up on Kubernetes — pick and deploy recipes, start router modes, validate disagg NIXL/UCX/NCCL interconnect, and triage day-2 failures.dynamo-interconnect-check, dynamo-recipe-runner, dynamo-router-starter, dynamo-troubleshoot
Earth2StudioOpen-source deep-learning framework for exploring, building and deploying AI weather/climate workflows.earth2studio-create-datasource, earth2studio-create-diagnostic, earth2studio-create-prognostic, earth2studio-data-fetch, earth2studio-deterministic-forecast, earth2studio-discover, earth2studio-install
HoloHubBuild, run, debug, benchmark, and develop HoloHub applications and Holoscan Modules with validated lifecycle workflows.holohub-app-lifecycle, holohub-debug-build-run, holohub-module-lifecycle
Holoscan SDKInstall and set up the Holoscan SDK on any platform (container, Debian, Python, Conda, or source).holoscan-install-debian, holoscan-install-source, holoscan-install-wheel, holoscan-install-conda, holoscan-install-container, holoscan-setup
Holoscan Sensor BridgeAgent-ready skills for Holoscan Sensor Bridge devkit workflows, including demo environment bring-up, FPGA flashing for Lattice and VB1940 hardware, example application execution, QA test-plan automation, and support for configuring and using the Holoscan Sensor Bridge FPGA intellectual property (IP) core.hsb-setup, hsb-flash, hsb-app, hsb-test, hsb-ip-def, hsb-ip-packetizer, hsb-ip-create-top
Isaac for Healthcare WorkflowsAgent-ready skills for Isaac for Healthcare agentic and catheter-navigation workflows, covering task authoring, data pipelines, policy training and validation, CT-derived digital twins, DRR rendering, and interactive catheter simulation.i4h-workflow, i4h-workflow-setup, i4h-workflow-create, i4h-workflow-scene-edit, i4h-workflow-dataset-teleop, i4h-workflow-dataset-replay, i4h-workflow-dataset-mimic, i4h-workflow-dataset-annotate, i4h-workflow-dataset-convert, i4h-workflow-finetune, i4h-workflow-validate, i4h-workflow-e2e, i4h-lerobot-viz, i4h-catheter-navigation, i4h-catheter-navigation-setup, i4h-catheter-navigation-digital-twin, i4h-catheter-navigation-render-drr, i4h-catheter-navigation-viewport, i4h-catheter-navigation-smoke, i4h-catheter-navigation-e2e
Jetson BSPAgentic skills for setting up and customizing an NVIDIA Jetson Linux Board Support Package (BSP) — pick a target, prepare image and sources, customize IO (camera, PCIe, USB, pinmux, clocks, and more), then promote, flash, and validate.jetson-build-source, jetson-customize-camera, jetson-customize-clocks, jetson-customize-fan, jetson-customize-mgbe, jetson-customize-nvpmodel, jetson-customize-pcie, jetson-customize-pinmux, jetson-customize-uphy, jetson-customize-usb, jetson-derive-carrier, jetson-download-bsp, jetson-flash-image, jetson-generate-kb, jetson-init-image, jetson-init-source, jetson-init-target, jetson-link-docs, jetson-optimize-memory, jetson-print-bsp-info, jetson-promote-image, jetson-quick-start, jetson-set-target, jetson-validate-image
Jetson DeviceDevice-side agent skills for working with a live NVIDIA Jetson after boot — diagnostics, memory auditing, headless setup, inference memory tuning, LLM serving and benchmarking, packaging guidance, and speculative decoding.jetson-diagnostic, jetson-headless-mode, jetson-inference-mem-tune, jetson-llm-benchmark, jetson-llm-serve, jetson-memory-audit, jetson-package, jetson-print-device-info, jetson-speculative-decoding
Medical AI SkillsAgent-ready medical AI skills built on MONAI for DICOM handling, NVIDIA-hosted medical imaging model workflows, segmentation, synthesis, and evidence-oriented evaluation.dicom-metadata-extract, dicom-series-preflight, dicom-series-to-volume, nv-generate-ct-rflow, nv-generate-mr, nv-generate-mr-brain, nv-generate-mr-brain-finetune, nv-generate-vae-finetune, nv-reason-cxr, nv-segment-ct, nv-segment-ct-finetune, nv-segment-ctmr
Megatron-CoreLarge-scale distributed training — model parallelism, pipeline parallelism, and mixed precision.mcore-create-issue, mcore-linting-and-formatting, mcore-run-on-slurm, mcore-split-pr, mcore-testing
NeMo AutoModelNeMo AutoModel - PyTorch-native distributed training for LLMs/VLMs with Hugging Face support, recipes, launchers, and validation workflows.nemo-automodel-distributed-training, nemo-automodel-launcher-config, nemo-automodel-model-onboarding, nemo-automodel-recipe-development
NeMo MBridgeNeMo MBridge - PyTorch-native bridge between Hugging Face and Megatron-Core for checkpoint conversion, training recipes, and NVIDIA GPU performance workflows.nemo-mbridge-mlm-bridge-training, nemo-mbridge-multi-node-slurm, nemo-mbridge-perf-activation-recompute, nemo-mbridge-perf-cpu-offloading, nemo-mbridge-perf-cuda-graphs, nemo-mbridge-perf-expert-parallel-overlap, nemo-mbridge-perf-hierarchical-context-parallel, nemo-mbridge-perf-megatron-fsdp, nemo-mbridge-perf-memory-tuning, nemo-mbridge-perf-moe-comm-overlap, nemo-mbridge-perf-moe-dispatcher-selection, nemo-mbridge-perf-moe-hardware-configs, nemo-mbridge-perf-moe-long-context, nemo-mbridge-perf-moe-optimization-workflow, nemo-mbridge-perf-moe-vlm-training, nemo-mbridge-perf-parallelism-strategies, nemo-mbridge-perf-sequence-packing, nemo-mbridge-perf-tp-dp-comm-overlap, nemo-mbridge-recipe-recommender, nemo-mbridge-resiliency
NeMo PlatformNeMo Platform brings NVIDIA NeMo libraries together under one CLI, Python SDK, and web UInemo-evaluator-plugin, nemo-data-designer-plugin
NeMo RelaySkills to help get started and use NeMo Relay - a runtime for instrumenting and controlling AI agents across harnesses, applications, and frameworks.nemo-relay-install, nemo-relay-get-started, nemo-relay-instrument-calls, nemo-relay-instrument-context-isolation, nemo-relay-instrument-typed-wrappers, nemo-relay-plugin-adaptive-tuning, nemo-relay-plugin-build, nemo-relay-plugin-observability, nemo-relay-migrate-from-flow, nemo-relay-debug-runtime-integration
NeMo RetrieverNeMo Retriever - deploy NeMo Retriever Library locally, extract information from corpus of data, and answer questions against the corpus.nemo-retriever
NeMo-RLRLHF training on Ray — GRPO, DPO, and SFT for LLMs and VLMs with FSDP2 and Megatron-Core.launch-nemo-rl, nemo-rl-auto-research, nemo-rl-brev-etiquette, nemo-rl-docs, nemo-rl-session-memory
NemoClawSecure agent sandboxing — run OpenClaw inside NVIDIA OpenShell with managed inference, policy management, remote deployment, sandbox monitoring.nemoclaw-user-guide
NemotronAuthor end-to-end model development, customization, evaluation, and deployment pipelines using the NVIDIA AI stack.nemotron-customize, nemotron-retrieval-recipes, nemotron-policy-generator
Nemotron SpeechDeploy and operate NVIDIA Nemotron Speech (Riva) NIMs — ASR, TTS, and NMT, cloud-hosted via build.nvidia.com or self-hosted on your own GPU.nemotron-speech, nemotron-asr-finetune
Physical AIPhysical AI skills for simulation, synthetic data generation, training, validation and deployment and more.omniverse-cad-to-simready, omniverse-realtime-viewer, omniverse-usd-performance-tuning, physical-ai-infrastructure-setup-and-resilient-scaling, physical-ai-neural-reconstruction, physical-ai-defect-image-generation, physical-ai-video-data-augmentation, physical-ai-people-attribute-search
PhysicsNeMoNVIDIA PhysicsNeMo - Open-source deep-learning framework for building, training, and fine-tuning deep learning models using state-of-the-art Physics-ML methods.physicsnemo-discover, physicsnemo-shard-tensor
Portfolio OptimizationGPU-accelerated Mean-CVaR portfolio optimization with NVIDIA cuOpt — CVaR optimization, efficient frontier, scenario generation, backtesting, and rebalancing.portfolio-optimization
RAG BlueprintRAG pipeline — deploy, configure, troubleshoot, and manage retrieval augmented generation with Docker Compose or Helm.rag-blueprint, rag-eval, rag-perf
Skill Card GeneratorReads an agent skill's source files and produces a skill card plus a review table. Use when a skill directory exists and a governance card needs to be generated or updated.skill-card-generator
TAO ToolkitNVIDIA TAO Toolkit - fine-tune and optimize 100+ pretrained vision AI models with your own data using low-code microservices, then export production-ready models for edge or cloud deployment.tao-analyze-changenet-rca, tao-finetune-huggingface-model, tao-port-huggingface-model, tao-run-automl, tao-run-automl-deft-pipeline, tao-run-deft-aoi, tao-run-inference-service, tao-train-single-step, paidf-anomalygen, tao-analyze-gaps-visual-changenet, tao-analyze-gaps-vlm-bcq, tao-convert-dataset-format, tao-generate-image-grounding, tao-generate-referring-expressions, tao-generate-video-reasoning-annotations, tao-mine-aoi-images, tao-route-visual-changenet-samples, tao-validate-dataset-format, tao-finetune-clip, tao-finetune-cosmos-embed, tao-finetune-cosmos-reason, tao-train-action-recognition, tao-train-bevfusion, tao-train-centerpose, tao-train-deformable-detr, tao-train-depth-anything-v2, tao-train-dino, tao-train-fast-foundation-stereo, tao-train-foundation-stereo, tao-train-grounding-dino, tao-train-image-classification, tao-train-mask-auto-encoder, tao-train-mask-auto-label, tao-train-mask-grounding-dino, tao-train-mask2former, tao-train-metric-learning-recognition, tao-train-nvdinov2, tao-train-nvpanoptix3d, tao-train-ocdnet, tao-train-ocrnet, tao-train-oneformer, tao-train-optical-inspection, tao-train-pointpillars, tao-train-pose-classification, tao-train-reid, tao-train-rtdetr, tao-train-segformer, tao-train-sparse4d, tao-train-visual-changenet, tao-run-on-brev, tao-run-on-docker, tao-run-on-kubernetes, tao-run-on-local-docker, tao-run-on-slurm, tao-run-platform, tao-setup-nvidia-gpu-host, tao-launch-workflow, tao-list-capabilities
TileGymTile-based GPU programming — adding new kernels, cross-framework conversion, and performance optimization.tilegym-adding-cutile-kernel, tilegym-converting-cutile-to-julia, tilegym-converting-cutile-to-triton, tilegym-cutile-autotuning, tilegym-cutile-python, tilegym-improve-cutile-kernel-perf, tilegym-monkey-patch-kernels-to-transformers
Video Search and SummarizationVSS Blueprint — deploy profiles, search and summarize video, generate analysis reports, manage alerts and incidents, query VIOS sensors, and use the RTVI VLM microservice.vss-ask-video, vss-deploy-dense-captioning, vss-deploy-detection-tracking-2d, vss-deploy-detection-tracking-3d, vss-deploy-profile, vss-deploy-video-embedding, vss-generate-video-calibration, vss-generate-video-report, vss-manage-alerts, vss-manage-video-io-storage, vss-query-analytics, vss-search-archive, vss-setup-behavior-analytics, vss-setup-video-analytics-api, vss-summarize-video

Getting Help & Contributing

Where to file an issue depends on what's broken:

  • Skill content issues (a specific skill has a bug, missing functionality, or incorrect content) — file in the source repo for that product, using the per-product table below.
  • Catalog issues (catalog README errors, sync workflow problems, distribution channels, signing/verification flow, docs in this repo) — file here using the catalog issue templates: Bug Report, Feature Request, or Documentation Request or Correction.
  • Questions or general discussion — use Discussions. The issue tracker is reserved for bug reports, feature proposals with a design, and documentation issues.
  • Security vulnerabilities — follow the disclosure process in SECURITY.md; do not open a public issue.

Per-product source repo links:

ProductIssuesDiscussionsContributingSecurity
AIQIssuesDiscussionsContributingSecurity
CUDA-QIssuesDiscussionsContributingSecurity
cuDFIssuesDiscussionsContributingSecurity
cuOptIssuesDiscussionsContributingSecurity
cuPyNumericIssues—Contributing—
DALIIssues—Contributing—
Data DesignerIssuesDiscussionsContributingSecurity
DeepStreamIssues—ContributingSecurity
Digital HealthIssues—ContributingSecurity
DOCAIssues—ContributingSecurity
DynamoIssuesDiscussionsContributingSecurity
Earth2StudioIssuesDiscussionsContributing—
HoloHubIssues—ContributingSecurity
Holoscan SDKIssues—ContributingSecurity
Holoscan Sensor BridgeIssues—Contributing—
Isaac for Healthcare WorkflowsIssues—ContributingSecurity
Jetson BSPIssues—ContributingSecurity
Jetson DeviceIssues—ContributingSecurity
Medical AI SkillsIssues—ContributingSecurity
Megatron-CoreIssuesDiscussionsContributing—
NeMo AutoModelIssuesDiscussionsContributingSecurity
NeMo MBridgeIssuesDiscussionsContributingSecurity
NeMo PlatformIssuesDiscussionsContributingSecurity
NeMo RelayIssuesDiscussionsContributingSecurity
NeMo RetrieverIssuesDiscussionsContributingSecurity
NeMo-RLIssuesDiscussionsContributingSecurity
NemoClawIssuesDiscussionsContributingSecurity
NemotronIssuesDiscussionsContributingSecurity
Nemotron SpeechIssues—ContributingSecurity
Physical AIIssues—ContributingSecurity
PhysicsNeMoIssuesDiscussionsContributingSecurity
Portfolio OptimizationIssuesDiscussionsContributingSecurity
RAG BlueprintIssuesDiscussionsContributingSecurity
Skill Card GeneratorIssues—ContributingSecurity
TAO ToolkitIssuesDiscussionsContributingSecurity
TileGymIssues—ContributingSecurity
Video Search and SummarizationIssuesDiscussionsContributingSecurity

For issues with this catalog repo itself (README, structure, listing a new product): open an issue here.


Verifying Skills

Every published skill ships with a detached OMS signature (skill.oms.sig). The sync pipeline drops any skill missing the required artifacts before publishing, so every skill in the catalog carries:

  • SKILL.md — the skill instructions consumed by the agent
  • skill-card.md — skill identity and governance card
  • skill.oms.sig — detached OMS signature (verifiable against nv-agent-root-cert.pem)
  • A Tier-3 evaluation dataset — accepted at evals/evals.json, evals/*.json, eval/*.json, or benchmark/evals.json
  • BENCHMARK.md — generated benchmark report capturing verifiable uplift data

Verify a skill against the NVIDIA trust anchor nv-agent-root-cert.pem:

pip install model-signing
model_signing verify certificate SKILL_DIR \
  --signature SKILL_DIR/skill.oms.sig \
  --certificate_chain nv-agent-root-cert.pem \
  --ignore_unsigned_files

A successful verification confirms that the skill contents have not been modified since signing by NVIDIA.

See Verify Signed Agent Skills for signature layout, the trust pipeline, and policy options.


Roadmap

  • ✅ Public skills catalog with NVIDIA-verified skills across multiple products
  • ✅ Automated sync pipeline with skills mirrored from product repos daily
  • ✅ Security scanning for all published skills covering instruction safety and supply-chain integrity
  • ✅ Skills signing so every published skill carries a verifiable NVIDIA signature
  • ✅ Skills universal evaluation criteria and task-specific criteria
  • ✅ Skill Card with machine-readable metadata for identity, provenance, quality, and behavioral boundaries
  • ✅ Sync-time compliance gates — signature drift detection and missing-artifact enforcement
  • ✅ Syndication to external marketplaces — Skills.sh, Codex plugin, Claude Code plugin, ClawHub, Hermes Hub
  • 🔲 Syndication to additional MCP hubs and partner channels

Repository Structure

NVIDIA/skills/
├── skills/                      # NVIDIA-verified skills (count grows continuously),
│   │                              synced from upstream product repos
│   ├── README.md                 # Browser-facing install guidance
│   ├── <product-prefix>-*/       # Flat layout — one dir per skill, product-prefixed
│   │                               # e.g. aiq-*, cuopt-*, cupynumeric-*,
│   │                               # dali-*, deepstream-*, dicom-*, digital-health-*,
│   │                               # dynamo-*, earth2studio-*, holoscan-*, hsb-*,
│   │                               # jetson-*, launch-nemo-rl, mcore-*,
│   │                               # nemo-automodel-*, nemo-data-designer-plugin,
│   │                               # nemo-evaluator-plugin, nemo-mbridge-* (20 skills),
│   │                               # nemo-retriever, nemo-rl-* (4 skills),
│   │                               # nemoclaw-user-guide, nemotron-*, nemotron-speech,
│   │                               # nv-* (medical AI), physicsnemo-*, rag-*,
│   │                               # skill-card-generator, tao-*, tilegym-*,
│   │                               # vss-* (15 skills), accelerated-computing-cudf,
│   │                               # cudaq-guide, portfolio-optimization
│   ├── omniverse-*/              # Physical AI — manually staged (see manual-components.yml)
│   └── physical-ai-*/            # Physical AI — manually staged
├── components.d/                # Product registry — one file per component, teams onboard here
│   ├── README.md                 # Schema and onboarding instructions
│   └── <product>.yml             # one file per registered product
├── plugins/                     # Packaged plugin distributions
│   └── nvidia-skills/            # Curated NVIDIA skills bundle (Claude Code, Codex)
├── plugins.d/                   # Plugin build registry — config for `build-plugins.py`
│   ├── README.md
│   ├── _defaults.yml
│   └── nvidia-skills.yml
├── .claude-plugin/              # Claude Code marketplace metadata
│   └── marketplace.json
├── .agents/plugins/             # Agent marketplace metadata (other clients)
│   └── marketplace.json
├── docs/                        # Long-form documentation (published via Fern)
│   ├── README.md                 # How to build the docs locally
│   ├── index.mdx
│   ├── advanced-install.mdx
│   ├── agent-skill-trust-pipeline.mdx
│   ├── release-checklist.mdx
│   ├── scanning-agent-skills.mdx
│   ├── signing-agent-skills.mdx
│   └── skill-cards.mdx
├── fern/                        # Fern docs site configuration
├── .github/
│   ├── workflows/                # Sync pipeline, plugin validation, DCO check, author verify
│   └── scripts/                  # regenerate-readme.sh, build-plugins.py,
│                                 # manual-components.yml (temp Physical AI catalog
│                                 # exception, removed after Computex 2026),
│                                 # marketplace/metadata.json (skill metadata sidecar)
├── nv-agent-root-cert.pem       # Trust anchor for OMS signature verification
├── skills.sh.json               # Skills.sh marketplace grouping config
├── CHANGELOG.md
├── CONTRIBUTING.md              # Contribution guidelines
├── SECURITY.md                  # Security reporting policy
├── CODE_OF_CONDUCT.md           # Community code of conduct
├── LICENSE-APACHE               # Apache 2.0 (source code)
└── LICENSE-CC-BY-4.0            # CC BY 4.0 (documentation/skills)

Skills are maintained in their respective product repos (see the Source column in the Skill Catalog) and synced to this repo daily. Products only appear under skills/ after the sync pipeline confirms each skill carries:

  • skill.oms.sig — detached OMS-format signature (verifiable against nv-agent-root-cert.pem)
  • skill-card.md — skill identity and governance card
  • A Tier-3 evaluation dataset — accepted at evals/evals.json, evals/*.json, eval/*.json, or benchmark/evals.json

When evaluation runs produce a BENCHMARK.md, it ships alongside the skill so consumers can see verifiable benchmark uplift data.


Standards & Compatibility

This repository adheres to the Agent Skills specification:

  • Skills are portable directories with a SKILL.md file at their root.
  • Metadata uses YAML frontmatter with required name and description fields.
  • Skills follow a progressive disclosure model — lightweight metadata loads at startup, full instructions load on activation.
  • Validate your skill using the skills-ref reference library.

License

Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.

This code is dual-licensed with documentation/skills under the CC-BY-4.0 AND source code under Apache-2.0 license terms. The full license texts can be found in LICENSE-APACHE and LICENSE-CC-BY-4.0 respectively.

其他

中风险

  • 来源需自行核对维护者身份。
  • 包含脚本或命令调用,安装前请复核。
  • 可能需要外部 token、网络权限或第三方服务。
  • 未检测到高风险命令。
  • 扫描发现:1 条。

Codex — Git Clone 安装

  1. 安装前请先查看来源仓库和风险报告。
  2. 克隆仓库:git clone https://github.com/NVIDIA/skills.git
  3. 将 "skills/doca-dpa" 文件夹复制到 Codex 的 skills 目录中。
  4. 重启 Codex 让新的 skill 生效。

Codex — 手动复制安装

  1. 安装前请先查看来源仓库和风险报告。
  2. 从源仓库下载 SKILL.md 及相关文件。
  3. 在 Codex 的 skills 目录中创建新文件夹。
  4. 将所有 skill 文件复制到新文件夹中。
  5. 重启 Codex 让新的 skill 生效。

Claude Code — Git Clone 安装

  1. 安装前请先查看来源仓库和风险报告。
  2. 克隆仓库:git clone https://github.com/NVIDIA/skills.git
  3. 将 "skills/doca-dpa" 文件夹复制到 Claude Code 的 skills 目录中。
  4. 重启 Claude Code 让新的 skill 生效。

Claude Code — 手动复制安装

  1. 安装前请先查看来源仓库和风险报告。
  2. 从源仓库下载 SKILL.md 及相关文件。
  3. 在 Claude Code 的 skills 目录中创建新文件夹。
  4. 将所有 skill 文件复制到新文件夹中。
  5. 重启 Claude Code 让新的 skill 生效。

Cursor — Git Clone 安装

  1. 安装前请先查看来源仓库和风险报告。
  2. 克隆仓库:git clone https://github.com/NVIDIA/skills.git
  3. 将 "skills/doca-dpa" 文件夹复制到 Cursor 的 skills 目录中。
  4. 重启 Cursor 让新的 skill 生效。

Cursor — 手动复制安装

  1. 安装前请先查看来源仓库和风险报告。
  2. 从源仓库下载 SKILL.md 及相关文件。
  3. 在 Cursor 的 skills 目录中创建新文件夹。
  4. 将所有 skill 文件复制到新文件夹中。
  5. 重启 Cursor 让新的 skill 生效。

GitHub Copilot — Git Clone 安装

  1. 安装前请先查看来源仓库和风险报告。
  2. 克隆仓库:git clone https://github.com/NVIDIA/skills.git
  3. 将 "skills/doca-dpa" 文件夹复制到 GitHub Copilot 的 skills 目录中。
  4. 重启 GitHub Copilot 让新的 skill 生效。

GitHub Copilot — 手动复制安装

  1. 安装前请先查看来源仓库和风险报告。
  2. 从源仓库下载 SKILL.md 及相关文件。
  3. 在 GitHub Copilot 的 skills 目录中创建新文件夹。
  4. 将所有 skill 文件复制到新文件夹中。
  5. 重启 GitHub Copilot 让新的 skill 生效。

Windsurf — Git Clone 安装

  1. 安装前请先查看来源仓库和风险报告。
  2. 克隆仓库:git clone https://github.com/NVIDIA/skills.git
  3. 将 "skills/doca-dpa" 文件夹复制到 Windsurf 的 skills 目录中。
  4. 重启 Windsurf 让新的 skill 生效。

Windsurf — 手动复制安装

  1. 安装前请先查看来源仓库和风险报告。
  2. 从源仓库下载 SKILL.md 及相关文件。
  3. 在 Windsurf 的 skills 目录中创建新文件夹。
  4. 将所有 skill 文件复制到新文件夹中。
  5. 重启 Windsurf 让新的 skill 生效。
查看 SKILL.md 原文
license: Apache-2.0
name: doca-dpa
description: >
  Use this skill when the user is doing hands-on DOCA DPA host-side
  work on a BlueField — creating the `doca_dpa` Core context, loading
  a DPACC-compiled DPA app image (`doca_dpa_app`), creating DPA
  threads, launching kernels via `doca_dpa_kernel_launch_update_*`,
  draining `doca_dpa_completion`, running `doca_dpa_cap_*` discovery,
  choosing between the DPA comm component (inter-DPA messaging) and the
  DPA verbs component (in-kernel RDMA), or debugging `DOCA_ERROR_*`
  from `doca_dpa_*`. Trigger even without "DOCA DPA" or "Data-Path
  Accelerator": "run compute on the DPA from my host", "DPA kernel
  hangs, no completion", "DOCA_ERROR_DRIVER on launch", "DOCA/DPACC
  version skew", or "does this BlueField expose a DPA". Route elsewhere
  for DPA-side kernel programming itself, DPACC compiler internals,
  host↔DPU messaging (doca-comch), host-side RDMA (doca-rdma), and
  GPU-initiated networking (doca-gpunetio).
metadata:
  kind: library
compatibility: >
  Requires DOCA SDK installed at /opt/mellanox/doca on Linux
  (Ubuntu 22.04/24.04 or RHEL/SLES) with a BlueField DPU whose
  generation exposes the DPA processor to the host, plus the
  matching DPACC compiler at a version listed compatible by
  the DOCA Compatibility Policy. Reads `pkg-config --modversion
  doca-dpa` and the installed `dpacc` version, and inspects
  /opt/mellanox/doca/{lib,include,samples/doca_dpa,applications}.

DOCA DPA

Where to start: This skill assumes DOCA is already installed, the user's BlueField has a DPA processor and the host can see it through DOCA, and the user is doing hands-on DPA work from the host side — i.e. using doca-dpa to load a DPA application image, launch DPA kernels, and exchange data with the DPA processor. Open TASKS.md if the user wants to do something (configure / build / modify / run / test / debug); open CAPABILITIES.md when the question is what can the host-side DPA API express on this version + this BlueField generation. If the user has not installed DOCA yet, route to doca-setup first; if the user is asking how to write the DPA-side kernel itself (the code that runs on the DPA processor, compiled by dpacc), that is a different scope — route via doca-public-knowledge-map to the public DOCA DPA / DPACC / DPA-Comms / DPA-Verbs guides (this skill does not redefine those DPA-side surfaces).

Example questions this skill answers well

The CLASSES of DPA questions this skill is built to answer, each with one worked example. The agent should treat the class as the load-bearing piece — the worked example is a single instance.

  • "How do I run a piece of compute on the DPA processor from my host program?" — worked example: "load a DPA kernel that counts events in a loop and reports the count back to the host". Answered by the two-side-program model + the host-side launch workflow in CAPABILITIES.md ## Capabilities and modes
  • "Does this BlueField even have a DPA, and which DPA features does my DOCA install expose?" — worked example: "my host has a BlueField; can I use the DPA on it for a programmable control workload?". Answered by the dual-axis capability rule (BlueField-generation axis via doca_dpa_cap_* against the active doca_devinfo plus the DOCA-install axis via pkg-config --modversion doca-dpa) in CAPABILITIES.md ## Capabilities and modes
  • "Why does my host-side DPA setup fail with DOCA_ERROR_NOT_SUPPORTED even though DOCA Core looks healthy?" — worked example: "the BlueField generation in this host predates the DPA feature my code uses". Answered by the env-precondition matrix in CAPABILITIES.md ## Safety policy
  • "How do I get arguments and results between my host program and my DPA kernel?" — worked example: "pass a buffer pointer and a length into the DPA kernel as launch arguments; read a completion back when the kernel finishes". Answered by the launch-argument + completion overlay in CAPABILITIES.md ## Capabilities and modes
  • "Is the DPA host-side API I'm reading about on my installed DOCA?" — worked example: "is the host-side launch helper I see in the docs available against the DOCA + DPACC versions on this host?". Answered by the version-compatibility overlay in CAPABILITIES.md ## Version compatibility which cross-links the canonical detection chain in doca-version and adds the DPA-specific DOCA must match DPACC overlay.
  • "What does this DOCA_ERROR_* from a doca_dpa_* call mean and which layer caused it?" — worked example: "DOCA_ERROR_DRIVER on a host-side launch call — is it DOCA, the DPACC-produced image, or the DPA processor itself?". Answered by the DPA overlay on the cross-library taxonomy in CAPABILITIES.md ## Error taxonomy
  • "How does my DPA kernel send a small message to another DPA thread on the same DPA processor?" — worked example: "two DPA threads in the same loaded doca_dpa_app; thread A sends a counter value to thread B over a DPA-side comms endpoint and thread B signals the host through doca_dpa_completion". Answered by the DPA-Comms routing rule, primitive families, host-side capability-budget rule, and DPA-Comms error overlay in CAPABILITIES.md ## comms plus the configure / build / modify / run / test / debug overlay in TASKS.md ## comms. Disambiguates the DPA device-side comm component from host-side doca-comch and host-side doca-rdma.
  • "How do I do RDMA directly from inside my DPA kernel to a remote peer, without round-tripping to the host?" — worked example: "my DPA kernel needs to fetch the next 4 KiB input buffer from a remote node via RDMA read before continuing compute; the host round-trip is the measured bottleneck". Answered by the 4-way RDMA matrix, the host-configures-QP / DPA-uses-QP coupling rule, the host-side cap-query rule for the specific verb, and the DPA-Verbs error overlay in CAPABILITIES.md ## verbs plus the workflow overlay in TASKS.md ## verbs. Includes the climb-back rule for when the latency-tuning premise stops holding.

Audience

This skill serves external developers building applications that consume the DOCA DPA library from the host side — i.e., users whose code calls doca_dpa_* from host C / C++ to stand up the per-DPA-instance context, load a DPA application image that dpacc produced from their DPA-side source, create one or more DPA threads, launch DPA kernels with arguments, and drain completions. It is not for NVIDIA developers contributing to DOCA DPA itself, nor is it the place to learn how to write the DPA-side kernel code (that path goes through the public DOCA DPA, DPACC, DPA-Comms, and DPA-Verbs guides via doca-public-knowledge-map).

Language scope. DOCA DPA ships as a host-side C library with pkg-config module name doca-dpa. The host-side API is C; the DPA-side kernel is a separate translation unit written in the language the DPACC compiler accepts and compiled by dpacc into a binary that the host packages into the executable as the DPA application image. The shipped samples under /opt/mellanox/doca/samples/doca_dpa/ are written in C plus DPA-side source (NVIDIA's choice). Other-language consumers are limited in practice — the DPA-side kernel has no FFI escape hatch because it must be a translation unit dpacc accepts — but a Rust / Go / Python host-side wrapper that drives doca_dpa_* setup and launches a DPA kernel image built separately is still useful, and the skill keeps the lifecycle, capability-discovery, env-precondition, and error-taxonomy guidance language-neutral.

When to load this skill

Load this skill when the user is doing hands-on DOCA DPA work from the host side, in any host language plus a DPA-side translation unit built by dpacc. Concretely:

  • Initializing a doca_dpa against a doca_dev that maps to a BlueField with a DPA processor visible to the host.
  • Loading a DPA application image (doca_dpa_app) that dpacc produced from the user's DPA-side source, into the doca_dpa context.
  • Creating one or more DPA execution contexts (doca_dpa_thread) so DPA kernels have somewhere to run on the DPA processor.
  • Launching a DPA kernel function with arguments from the host via the doca_dpa_kernel_launch_update_* family, and reasoning about which argument shape is supported on this install.
  • Attaching a doca_dpa_completion to observe when async DPA work finishes, and draining it from the host side.
  • Checking which DPA features are supported on the active doca_devinfo via the doca_dpa_cap_* family — BlueField generations differ in DPA hardware support.
  • Debugging a DOCA_ERROR_* returned from a doca_dpa_* call — in particular disambiguating DPA not present on this BlueField from DPA feature too new for this hardware generation from DPACC-produced image mismatched against the host-side DOCA install from DPA driver layer reporting failure.
  • Designing host-side bindings in a non-C language that drive a DPA application image they built separately with dpacc — the env-precondition and capability-discovery rules in this skill still apply.
  • Writing DPA-side kernel code that calls the DPA device-side comm component libdoca_dpa_dev_comm.a (header doca_dpa_dev_comch_msgq.h) — for inter-DPA-thread messaging or coordination signals between DPA threads on the same doca_dpa_app. The DPA-Comms routing rule, primitive families, capability rule (there is no per-primitive host cap-query family — host-side DPA discovery is only doca_dpa_cap_is_supported / doca_dpa_cap_get_max_kernel_time_alive_supported), error overlay (_AGAIN → kernel must yield; _BAD_STATE disambiguation from the parent's host-side _BAD_STATE), and the configure / build / modify / run / test / debug overlay live in CAPABILITIES.md ## comms and TASKS.md ## comms under this same skill.
  • Writing DPA-side kernel code that calls the DPA device-side verbs component libdoca_dpa_dev_verbs.a (header doca_dpa_dev_verbs.h) — for RDMA from inside the DPA kernel to a remote peer when the host round-trip is the measured latency bottleneck. The 4-way RDMA matrix, the host-configures-QP / DPA-uses-QP coupling rule, the capability rule (no per-verb host cap-query family exists; verb availability follows the BlueField generation + matched DOCA/DPACC install, read from doca_dpa_dev_verbs.h and the shipped sample), the IO_FAILED → CQE-inspection overlay, and the climb-back rule live in CAPABILITIES.md ## verbs and TASKS.md ## verbs under this same skill.

Do not load this skill for general DOCA orientation, install of DOCA or the DPACC compiler, the DPA-side programming model itself (how to write a DPA kernel; the DPA device-side comm and verbs components (libdoca_dpa_dev_comm.a / libdoca_dpa_dev_verbs.a) that run inside the DPA kernel), or non-DPA library questions. For those, route through doca-public-knowledge-map to the matching upstream guide.

What this skill provides

This is a thin loader. The body keeps only the orientation needed to pick the right next file. The substantive DPA-specific material lives in two companion files:

  • CAPABILITIES.md — what the host-side DPA API can express on this version + this BlueField generation: the per-DPA-instance doca_dpa context, the loaded doca_dpa_app image produced by dpacc, the doca_dpa_thread execution context, the host-initiated kernel launch surface (doca_dpa_kernel_launch_update_*), the doca_dpa_completion mechanism, the capability-query surface (doca_dpa_cap_*), the DPA error taxonomy mapped onto the cross-library DOCA_ERROR_* set, the observability surface (host-side completions plus the public DPA developer tools surface reachable via doca-public-knowledge-map), and the safety policy that gates env preconditions (DPA-capable BlueField, matched DOCA + DPACC versions, DPA-side image and host-side expected entry points agree).
  • TASKS.md — step-by-step workflows for the six in-scope DPA verbs: configure, build, modify, run, test, debug. Plus a Deferred task verbs block that points out-of-scope questions at the right next skill.

The skill assumes a host where DOCA is already installed at the standard location, a BlueField with a DPA processor is physically present and visible to the host, the DPACC compiler is installed at a version matched to the DOCA install per the DOCA Compatibility Policy, and the user already knows how (at least at a sketch level) to write the DPA-side kernel that dpacc will compile. It does not cover installing DOCA or the DPACC compiler — that path goes through doca-setup.

What this skill deliberately does not ship

This skill is agent guidance, not a samples or templates bundle. To keep the boundary clean, it deliberately does not contain — and pull requests should not add:

  • Pre-written DOCA DPA application source code or DPA-side kernel source, in any language. The verified DPA source is the shipped C + DPA-side samples at /opt/mellanox/doca/samples/doca_dpa/. The agent's job is to route the user to those files and prescribe a minimum-diff modification on them via the universal modify-a-sample workflow in doca-programming-guide, layered with the DPA-specific overrides in TASKS.md ## modify.
  • Standalone build manifests (meson.build, CMakeLists.txt, …) parked inside the skill. The agent constructs the build manifest in the user's project directory against the user's installed DOCA + DPACC compiler, where pkg-config --modversion doca-dpa and the installed dpacc are the two sources of truth.
  • A samples/, bindings/, or reference/ subtree of any kind. A mock or incomplete artifact in this skill's tree, even one labeled "reference", is misleading: users will read it as buildable.
  • DPA device-side content for the comm / verbs components (libdoca_dpa_dev_comm.a / libdoca_dpa_dev_verbs.a). These are DPA-side archives shipped as part of doca-dpa (NOT separate pkg-config modules): their symbols are called from inside the DPA kernel and linked into the DPA image by dpacc, not from the host. Their public guides are reachable via doca-public-knowledge-map. This skill names them and routes; it does not redefine them.

Loading order

  1. Read this SKILL.md first to confirm the user's question is in scope (host-side DPA work, not DPA-side kernel-writing).
  2. For the DPA capability matrix, the doca_dpa per-instance context, the loaded doca_dpa_app image, the doca_dpa_thread execution context, the kernel-launch + completion model, the dual capability query, the env-precondition policy, the error taxonomy, the observability surface, and the safety policy, see CAPABILITIES.md.
  3. For step-by-step workflows — configure, build, modify, run, test, debug — see TASKS.md.

Both companion files cross-link to each other, doca-version for the canonical DOCA version-handling rules (with the DPA overlay that DOCA must match the DPACC compiler), and doca-public-knowledge-map whenever the right answer is "look it up in the public DOCA DPA, DPACC, DPA-Comms, or DPA-Verbs guide, or in the on-disk install layout" rather than "DPA host-side-specific guidance".

Related skills

  • doca-public-knowledge-map — the routing table for every public DOCA documentation source and the on-disk layout of an installed DOCA package. The DPA public guide is at https://docs.nvidia.com/doca/sdk/DOCA-DPA/index.html; the DPACC compiler guide, the DPA-Comms guide (DPA-side communications), the DPA-Verbs guide (DPA-side verbs), and the DPA Tools umbrella (developer / admin CLIs for DPA) live in the same routing table and are companion surfaces to this skill rather than redefined by it.
  • doca-setup — env preparation, install verification, DPACC compiler install / verification, and the I have no install yet path with the public NGC DOCA container. This skill assumes its preconditions are satisfied AND that DPACC is installed at a version that matches DOCA.
  • doca-version — canonical DOCA version-handling rules. This skill's ## Version compatibility cross-links the four-way match rule and adds the DPA-specific DOCA-and-DPACC must match overlay per the DOCA Compatibility Policy.
  • doca-structured-tools-contract — the bundle's structured-tools precedence rule (detect / prefer / fall back / report). The Command appendix in TASKS.md honors this contract.
  • doca-programming-guide — general DOCA programming patterns shared by every library: the canonical pkg-config + meson build pattern, the universal modify-a-shipped-sample first-app workflow, the universal Core-context lifecycle, the cross-library DOCA_ERROR_* taxonomy, and the program-side debug order. This skill layers DPA specifics on top.
  • doca-debug — the cross-cutting debug ladder (install / version / build / link / runtime / program / driver). DPA-specific debug (DPACC + DOCA version skew, DPA not present on this BlueField generation, DPA kernel hangs that show no host-side completion, launch-argument shape mismatches between the host launch call and the DPA-side function signature) overlays on top of that ladder.

DOCA DPA's DPA device-side components — the comm archive libdoca_dpa_dev_comm.a (communication primitives the DPA kernel itself calls, header doca_dpa_dev_comch_msgq.h) and the verbs archive libdoca_dpa_dev_verbs.a (ibverbs-like RDMA verbs the DPA kernel itself calls, header doca_dpa_dev_verbs.h) — are DPA-side archives shipped within doca-dpa, not separate pkg-config modules, and each has its own public guide. No library skill ships for them in this bundle yet; for any DPA-side question, route via doca-public-knowledge-map to the public DOCA DPA Comms and DOCA DPA Verbs guides and to the shipped /opt/mellanox/doca/samples/doca_dpa/ samples (which include both host-side and DPA-side translation units). Conflating them with doca-dpa is the single most common DPA first-app design error.

发现问题?提交给管理员复核

评分:

评论 (0)

暂无评论,成为第一个评论者吧!