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。
SkillAtlasSkill 详情
hsb-ip-def
Official, NVIDIA-verified Agent Skills for Claude Code, Codex, and other coding agents.
审核状态：已审核Quality 80Security 80
项目 README
来源文件：README.md
NVIDIA Agent Skills
Official, NVIDIA-verified Agent Skills for Claude Code, Codex, and other coding agents.
📖 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
skillsCLI (v1.5.16 or newer). Installing vianpx skills@latest add nvidia/skillsalways 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
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:
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 agentskill-card.md— skill identity and governance cardskill.oms.sig— detached OMS signature (verifiable againstnv-agent-root-cert.pem)- A Tier-3 evaluation dataset — accepted at
evals/evals.json,evals/*.json,eval/*.json, orbenchmark/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 againstnv-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, orbenchmark/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.mdfile at their root. - Metadata uses YAML frontmatter with required
nameanddescriptionfields. - Skills follow a progressive disclosure model — lightweight metadata loads at startup, full instructions load on activation.
- Validate your skill using the
skills-refreference 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。
- 未检测到明显外部权限要求。
- 未检测到高风险命令。
- 扫描发现：1 条。
Codex — Git Clone 安装
- 安装前请先查看来源仓库和风险报告。
- 克隆仓库：git clone https://github.com/NVIDIA/skills.git
- 将 "skills/hsb-ip-def" 文件夹复制到 Codex 的 skills 目录中。
- 重启 Codex 让新的 skill 生效。
Codex — 手动复制安装
- 安装前请先查看来源仓库和风险报告。
- 从源仓库下载 SKILL.md 及相关文件。
- 在 Codex 的 skills 目录中创建新文件夹。
- 将所有 skill 文件复制到新文件夹中。
- 重启 Codex 让新的 skill 生效。
Claude Code — Git Clone 安装
- 安装前请先查看来源仓库和风险报告。
- 克隆仓库：git clone https://github.com/NVIDIA/skills.git
- 将 "skills/hsb-ip-def" 文件夹复制到 Claude Code 的 skills 目录中。
- 重启 Claude Code 让新的 skill 生效。
Claude Code — 手动复制安装
- 安装前请先查看来源仓库和风险报告。
- 从源仓库下载 SKILL.md 及相关文件。
- 在 Claude Code 的 skills 目录中创建新文件夹。
- 将所有 skill 文件复制到新文件夹中。
- 重启 Claude Code 让新的 skill 生效。
Cursor — Git Clone 安装
- 安装前请先查看来源仓库和风险报告。
- 克隆仓库：git clone https://github.com/NVIDIA/skills.git
- 将 "skills/hsb-ip-def" 文件夹复制到 Cursor 的 skills 目录中。
- 重启 Cursor 让新的 skill 生效。
Cursor — 手动复制安装
- 安装前请先查看来源仓库和风险报告。
- 从源仓库下载 SKILL.md 及相关文件。
- 在 Cursor 的 skills 目录中创建新文件夹。
- 将所有 skill 文件复制到新文件夹中。
- 重启 Cursor 让新的 skill 生效。
GitHub Copilot — Git Clone 安装
- 安装前请先查看来源仓库和风险报告。
- 克隆仓库：git clone https://github.com/NVIDIA/skills.git
- 将 "skills/hsb-ip-def" 文件夹复制到 GitHub Copilot 的 skills 目录中。
- 重启 GitHub Copilot 让新的 skill 生效。
GitHub Copilot — 手动复制安装
- 安装前请先查看来源仓库和风险报告。
- 从源仓库下载 SKILL.md 及相关文件。
- 在 GitHub Copilot 的 skills 目录中创建新文件夹。
- 将所有 skill 文件复制到新文件夹中。
- 重启 GitHub Copilot 让新的 skill 生效。
Windsurf — Git Clone 安装
- 安装前请先查看来源仓库和风险报告。
- 克隆仓库：git clone https://github.com/NVIDIA/skills.git
- 将 "skills/hsb-ip-def" 文件夹复制到 Windsurf 的 skills 目录中。
- 重启 Windsurf 让新的 skill 生效。
Windsurf — 手动复制安装
- 安装前请先查看来源仓库和风险报告。
- 从源仓库下载 SKILL.md 及相关文件。
- 在 Windsurf 的 skills 目录中创建新文件夹。
- 将所有 skill 文件复制到新文件夹中。
- 重启 Windsurf 让新的 skill 生效。
查看 SKILL.md 原文
name: hsb-ip-def
author: "Holoscan Team <holoscan-team@nvidia.com>"
description: Generate, validate, compare, or explain HSB HOLOLINK_def.svh macros. Do not use for FPGA_top.sv wrappers or packetizer-only derivation. Generation runs bundled Python scripts locally through shell commands and writes validated .svh output files after user-confirmed paths.
version: "0.1.0"
tags:
- holoscan
- hsb
- fpga
- systemverilog
- configuration
permissions: [file_read, file_write, shell]
license: Apache-2.0
compatibility: Targets HSB IP rev 16'h2604; backward-compatible with 16'h2603 (the public release rev). Prefer live HSB IP source when available; warn on unknown revisions. Requires Python 3.9+ with PyYAML for the generator.
metadata:
author: "Holoscan Team <holoscan-team@nvidia.com>"
team: holoscan
domain: fpga
vendor: nvidia
tags:
- holoscan
- hsb
- fpga
- systemverilog
- configuration
languages:
- systemverilog
- python
artifact: HOLOLINK_def.svh
hsb_ip_version: "16'h2604"
min_compat_rev: "16'h2603"HSB IP Def Skill
Purpose
Use this skill through four workflows:
- Generate a
HOLOLINK_def.svhfrom confirmed board requirements. - Validate an existing def file with the bundled validator.
- Explain / Reason about HSB IP macros, legality, and macro-driven ports.
- Compare two def files semantically.
Scope
This skill owns the contents of HOLOLINK_def.svh: \definedirectives,localparamarrays, theHOLOLINK_pkgwrapper, and boot-timeinit_reg[]sequence. The surrounding top-level wrapper is owned byhsb-ip-create-top; packetizer-only profile derivation can be delegated to hsb-ip-packetizer`.
The file must use the standard guard plus package HOLOLINK_pkg wrapper. Load references/macro-reference.md for the full wrapper requirement and per-macro semantics.
Prerequisites
- Python 3.9+ is required for the bundled scripts; PyYAML is required when reading YAML profiles.
- The user must confirm file writes and shell commands unless they already asked for that exact operation.
- A concrete def file path, pasted content, or confirmed generation profile is required before script-backed validation or generation.
- Live HSB IP source is optional but preferred when validating against a specific checked-out IP revision.
Instructions
- Run bundled script preflight from
references/script-usage.mdbefore Generate, Validate, Compare, or script-backed legality checks. - Never silently default a macro during Generate. Show the proposed value or inferred requirement and get user confirmation.
- Ask one requirements question per turn during chat-driven Generate.
- Use sensor-agnostic language unless the user says the design is camera-specific.
- Avoid unsupported "typical", "common", "most designs", or corpus-frequency claims. Anchor choices to IP behavior, documented constraints, or user requirements.
- Always surface HD-W3xx footgun warnings, even when the user asks only about errors.
- Do not generate
FPGA_top.sv; offer handoff tohsb-ip-create-topafter a generated def file is validated. - Treat unfamiliar macros as project-specific unless they are documented in this skill's references or live HSB IP source.
Security Considerations
This skill can read and write local files and run shell commands through its bundled scripts. Before running a command or writing a file, state the command or path and get user confirmation unless the user already explicitly requested that exact operation. For pasted HOLOLINK_def.svh content, write only to a safely generated file in an isolated temporary directory and remove it after validation unless the user asks to keep it.
Version, Compatibility, And Live Source
This skill targets HSB IP rev 16'h2604 and is backward-compatible with 16'h2603. Live HSB IP source supersedes bundled references.
When source-sensitive behavior matters:
- Locate
<hsb-ip-root>/top/HOLOLINK_top.sv. Known roots includehw/nvcpu_dgx_fpga/vrtl/hololink/and public-releasefpga/nv_hsb_ip/. - Read
HOLOLINK_REVandHOLOLINK_BACKWARD_COMPAT_REV. - Trust live source for consumed macros, port gates, and RTL behavior when it differs from bundled references.
- Pass
--ip-source <root>toscripts/validate_def.pywhen validating against a known source root.
Public-doc baseline:
https://github.com/nvidia-holoscan/holoscan-sensor-bridge/blob/release-2.6.0-EA/docs/user_guide/ip_integration.mdhttps://github.com/nvidia-holoscan/holoscan-sensor-bridge/blob/release-2.6.0-EA/docs/user_guide/port_description.md
Workflow Decision
- Generate when the user asks to create, scaffold, draft, design, or produce a
HOLOLINK_def.svh. - Validate when the user asks to lint, check, validate, or review a
HOLOLINK_def.svh. - Explain / Reason when the user asks what a macro does, whether a combination is legal, why validation fails, or how a macro affects
HOLOLINK_top. - Compare when the user asks to diff two def files or understand what changed between configurations.
Generate
Load references/generate-workflow.md and references/script-usage.md.
Follow the detailed Generate workflow in the reference: run preflight, classify supplied requirements, ask one requirements question per turn, build a flat YAML profile, run scripts/generate_def.py, show provenance for generated fields, and offer handoff to hsb-ip-create-top.
When packetizer fields are needed, invoke hsb-ip-packetizer with the known RX count, RX widths, and the user's data-manipulation description. Consume only that skill's packetizer_profile_overlay YAML keys, merge them into the in-progress profile, and continue full-file generation and validation here.
Validate
Load references/script-usage.md. Load references/validation-rules.md only when explaining specific rule IDs or validation behavior.
Steps:
- Run bundled script preflight once per session.
- Locate the file. If the user pastes content, tell them the generated temporary path before writing, ask for confirmation, use an isolated safe temp path, and clean it up after validation unless they ask to keep it. Otherwise use the provided path.
- Run
<PY> scripts/validate_def.py <path> --jsonand do not reimplement validation in context. - Group findings by severity: errors, warnings, then info. For each error, cite rule ID, line number, and macro when available.
- Always surface footgun warnings, especially HD-W3xx silent-fallback warnings.
- If clean, confirm the inferred archetype and IP version, then suggest the likely next check or handoff.
Explain / Reason
Load only the reference needed for the question:
| Question type | Reference |
|---|---|
| Macro semantics or legal values | references/macro-reference.md |
| Validation rule behavior | references/validation-rules.md |
| Macro-driven port effects | references/top-port-map.md |
init_reg[] and N_INIT_REG | references/init-reg-cookbook.md |
| Advanced macros | references/advanced-macros.md |
| Example legal configurations | references/archetypes.md |
For grounded legality questions, run preflight and prefer validating a concrete file or minimal synthetic file with scripts/validate_def.py over hand reasoning. Cite RTL line ranges from references when explaining why a rule exists.
Compare
Load references/script-usage.md, run preflight, then use <PY> scripts/compare_defs.py <a.svh> <b.svh> [--json|--text]. Summarize semantic differences, not whitespace or comment-only changes.
Limitations
- Do not generate
FPGA_top.sv; usehsb-ip-create-topafter the defs file validates. - Do not derive packetizer-only field sets here when the packetizer behavior is underspecified; delegate that slice to
hsb-ip-packetizer. - Do not treat bundled archetypes or corpus metadata as norms. They are examples and maintenance metadata, not defaults.
- Do not silently accept unknown macros as validated HSB IP behavior unless live source or references document them.
Troubleshooting
- Script preflight fails: report the missing Python or PyYAML requirement and stop before generation or validation.
- Validation reports errors: group by severity, cite rule ID and line, and fix the defs file before offering top-level handoff.
- Validation reports HD-W3xx warnings: surface them even when there are no errors because they describe silent RTL fallback risks.
- Unknown macro appears: treat it as project-specific unless live HSB IP source or bundled references document it.
Available Scripts
Use <PY> selected during preflight from references/script-usage.md for every command.
| Script | Purpose | Arguments |
|---|---|---|
scripts/generate_def.py | Generate HOLOLINK_def.svh from an archetype and/or YAML/JSON profile; validates before writing | --profile <path>, --archetype <slug>, -o <output>, optional --allow-random-uuid compatibility flag |
scripts/validate_def.py | Validate a HOLOLINK_def.svh and emit JSON or text findings | <path/to/HOLOLINK_def.svh>, optional --json or --text, optional --ip-source <root> |
scripts/compare_defs.py | Compare two def files semantically, ignoring whitespace/comment-only changes | <a.svh> <b.svh>, optional --json or --text |
scripts/build_corpus_metadata.py | Maintenance helper to rebuild anonymized corpus metadata; do not run during normal user workflows | <path1> [<path2> ...] |
Bundled Resources
| Resource | Use |
|---|---|
references/generate-workflow.md | Detailed Generate workflow, requirement order, question style, and per-topic prompt guidance |
references/script-usage.md | Preflight, command forms, and run_script() examples for bundled scripts |
references/macro-reference.md | Wrapper requirement, macro semantics, legal constraints, and RTL citations |
references/validation-rules.md | Rule catalog for validator findings |
references/archetypes.md | Illustrative legal configurations; never treat as templates or frequency guidance |
references/init-reg-cookbook.md | Boot-time APB write sequence patterns and address conventions |
references/top-port-map.md | Macro-to-HOLOLINK_top port effects |
references/advanced-macros.md | SYNC_CLK_HIF_APB, SYNC_CLK_HIF_PTP, PERI_RAM_DEPTH, and DISABLE_COE |
assets/metadata/corpus.json, assets/metadata/corpus-stats.json | Maintenance metadata only; do not cite corpus counts as user guidance |
scripts/generate_def.py | Generate a def file from a profile |
scripts/validate_def.py | Validate a def file and emit JSON/text findings |
scripts/compare_defs.py | Compare two def files semantically |
Examples
Use hsb-ip-def to generate a HOLOLINK_def.svh for a new HSB board.Treat as Generate, run script preflight, classify supplied requirements, ask one requirement question per turn, and runscripts/generate_def.pyonly after the profile is confirmed.Use hsb-ip-def to validate my existing HOLOLINK_def.svh and tell me whether any warnings are important.Treat as Validate, ask for or locate the file, runscripts/validate_def.py <path> --json, group findings by errors, warnings, and info, and always surface HD-W3xx footgun warnings.Use hsb-ip-def to explain whether HOST_WIDTH=512 and PTP_CLK_FREQ=90_000_000 is legal.Treat as Explain / Reason, prefer a concrete validator-backed check over hand reasoning, and load only the macro or validation reference needed to explain the result.
评论 (0)
暂无评论，成为第一个评论者吧！