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
doca-pcc-counters
Official, NVIDIA-verified Agent Skills for Claude Code, Codex, and other coding agents.
审核状态：已审核Quality 72Security 52
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
高风险
- 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。
- 存在潜在风险命令，请谨慎安装。
- 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/doca-pcc-counters" 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/doca-pcc-counters" 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/doca-pcc-counters" 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/doca-pcc-counters" 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/doca-pcc-counters" 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
license: Apache-2.0
name: doca-pcc-counters
description: >
Use this skill when the user is invoking the DOCA PCC Counters tool
— the `pcc_counters.sh` bash script under the DOCA tools directory —
to arm and read the fixed firmware/hardware PCC (Programmable
Congestion Control) diagnostic counters (CNP, RTT, WRED-drop, etc.)
on a ConnectX / BlueField device via mst + the mlx5 debugfs
`diag_cnt` interface. The script takes two positional args —
`set | query` and an mst device path — with no `--help` or
subcommands. Trigger even without "pcc_counters.sh" or "PCC
counters": "how do I read the CNP / RTT / WRED-drop counters",
"PCC counter stuck at zero", "the script says Bad Device", or "is
congestion control dropping packets on this port?". Route elsewhere
for writing a custom PCC algorithm (doca-pcc), factory firmware PCC
config, DOCA install, or fleet-wide CC tuning.
metadata:
kind: tool
compatibility: >
Requires DOCA/MFT on Linux with a ConnectX-6 or newer or BlueField
device, mst tools, mounted debugfs, and root access to the mlx5
diag_cnt interface. The script reads fixed firmware and hardware
diagnostics independently of custom PCC code. Its set operation
changes which counters are collected, not congestion-control or
forwarding behavior; treat later fleet tuning as a separate
high-stakes action.DOCA PCC Counters (pcc_counters.sh)
Where to start: This is a tool skill for invoking
pcc_counters.sh — a small bash script that arms and reads the
device's fixed set of firmware / hardware PCC diagnostic
counters (CNP count, RTT-perf, WRED-drop, RTT-gen, handled
events) through the mlx5 debugfs diag_cnt interface. Open
TASKS.md and start at ## run for
the canonical set-then-query sequence, or
## debug when the user reports
"ERROR: Bad Device", "counter stuck at zero", or "the
dump is empty". Open CAPABILITIES.md when
the question is which counters the script reports and how it
reaches them. If the user has not installed DOCA / MFT yet,
route to doca-setup first.
This skill is the firmware / HW PCC counter readout surface.
It is NOT the host-side control library that loads custom
congestion-control kernels onto the DPA (that is
doca-pcc) and it is NOT the
firmware PCC algorithm configuration (that path is firmware
configuration, routed via
doca-public-knowledge-map).
The counters this script reads are device / firmware
diagnostic counters that exist regardless of whether a custom
doca-pcc DPA kernel is running — do not condition them on a
custom kernel being loaded.
Example questions this skill answers well
The CLASSES of pcc_counters.sh questions this skill is built
to answer, each with one worked example. The class is the
load-bearing piece; the worked example is one instance.
- "How do I read the PCC diagnostic counters on this
device?" — worked example: "arm and dump the CNP / RTT /
WRED-drop counters for
/dev/mst/mt41692_pciconf0". Answered by the fixed counter set inCAPABILITIES.md ## Capabilities and modes- the
set-then-queryinvocation inTASKS.md ## run.
- the
- "What is the smallest legal invocation?" — worked
example: "what exactly do I type?". Answered by the
two-positional-argument contract
(
set | query+ an mst device path) inTASKS.md ## run. - "The script printed
ERROR: Bad Device— what's wrong?" — worked example: "my device path is not matching". Answered by the device-resolution layer inCAPABILITIES.md ## Error taxonomy - "A counter is stuck at zero — is the device idle, the
counters not armed, or genuinely no events?" — worked
example: "
PCC_CNP_COUNTreads 0 afterquery". Answered by the arm-before-read rule and the layered diagnosis inTASKS.md ## debug+CAPABILITIES.md ## Error taxonomy. - "Is this script on my install, and where?" — worked
example: "is
pcc_counters.shpresent and where does the install put it". Answered by the install overlay inCAPABILITIES.md ## Version compatibility, which redirects to the canonicaldoca-versionrules.
Audience
This skill serves operators, developers, and AI agents who need to read a ConnectX / BlueField device's firmware PCC diagnostic counters to reason about congestion-control behaviour (CNP generation, RTT requests/responses, WRED drops) on a port. Concretely:
- A network operator confirming whether congestion-control events (CNPs, RTT, WRED drops) are occurring on a device.
- A developer correlating a custom
doca-pccalgorithm's effect with the device-level PCC diagnostic counters (the script reads the firmware counters; the custom algorithm itself is a separate surface owned bydoca-pcc). - An AI agent producing a PCC counter snapshot as evidence for a congestion-control investigation.
It is not for users debugging the script's bash itself,
not the place to learn how to write a custom PCC
algorithm — that audience belongs in
doca-pcc — and not the
place for users who want to configure the factory firmware PCC
algorithm (route via
doca-public-knowledge-map).
pcc_counters.sh is shipped as a plain bash script
installed under the DOCA tools directory (per install_data in
tools/pcc_counters/meson.build), not a compiled binary and
not a library you link against. The skill uses the bundle's
kind: tool three-file shape (SKILL.md + CAPABILITIES.md
TASKS.md) so the agent's task-verb contract (configure / build / modify / run / test / debug) is uniform across the bundle.
When to load this skill
Load this skill when the user is — or the agent needs to — arm and read the device PCC diagnostic counters on a host or BlueField Arm with the mst tools available and debugfs mounted. Concretely:
- Arming the diagnostic counters with
seton a target mst device. - Reading the armed counters with
queryand quoting the named counter lines verbatim. - Capturing a counter readout as evidence for a congestion-control investigation.
Do not load this skill for general DOCA orientation,
custom-PCC algorithm design, the host-side doca-pcc library
API, the factory firmware PCC algorithm, or DOCA / MFT install.
For those, route to
doca-public-knowledge-map,
doca-pcc, or
doca-setup.
What this skill provides
This is a thin loader. Substantive material lives in two companion files:
CAPABILITIES.md— whatpcc_counters.shdoes: the exact two-operation surface (setarms the device's diagnostic counters by writing counter IDs + params to debugfs;queryreads thediag_cnt/dumpand prints the named counters), the FIXED firmware / HW counter set it knows (PCC_CNP_COUNT, theMAD_RTT_PERF_CONT_*, the*_EVENT_WRED_DROPfamily,HANDLED_*_EVENTS, theDROP_RTT_PORT*/RTT_GEN_PORT*families), how it resolves an mst device to a PCI address (mst status -v+lspci) and reaches/sys/kernel/debug/mlx5/<pci>/diag_cnt/, the install-availability overlay that redirects todoca-version, the layered error taxonomy (script-not-present / bad-device / not-armed-before-query / debugfs-or-permission / counter-stuck-at-zero / cross-cutting), and the safety policy that flagssetas a privileged debugfs write and any CC tuning decision derived from a reading as high-stakes.TASKS.md— step-by-step workflows for the in-scope task verbs:configure(route to install + confirm mst / debugfs / sudo),build(route to install; nothing to compile — it is a script),modify(refuse — do not patch the shipped script),run(theset-then-querysequence),test(confirm the dump contains the named counters with finite values),debug(the layered diagnosis ladder), plus aDeferred task verbsblock and aCommand appendix.
The skill assumes a host or BlueField where DOCA / MFT is
already installed (mst tools present, debugfs mounted, sudo
available) and the target device is visible to mst status -v.
What this skill deliberately does not ship
This skill is agent guidance, not a samples or scripts bundle. To keep the boundary clean, it deliberately does not contain — and pull requests should not add:
- Invented flags or subcommands.
pcc_counters.shhas exactly two operations (set,query), takes exactly two positional arguments, and has NO--help,--version,list,snapshot,watch, ordiff. Do not invent any. - Pre-baked example counter values. Counter values are device-, firmware-, and traffic-state-specific; a captured value will mislead an operator on a different device.
- Wrappers, parsers, or rewritten copies of the script. The script is the contract; modifying or re-implementing it is out of scope.
- A specific congestion-control tuning recommendation derived from a counter reading. That is a high-stakes domain question — the skill prescribes how to capture the counters; it refuses to translate a delta into a CC parameter change without the user's own domain analysis.
- A
samples/orreference/subtree. This is a thin loader for a documented script; substantive material lives in the script and the public PCC documentation.
Loading order
- Read this
SKILL.mdfirst to confirm the user's question is in scope (reading the device's firmware PCC diagnostic counters; not designing or loading a custom algorithm). - For what the script does, the fixed counter set, the debugfs mechanism, the install-availability overlay, the layered error surface, and the safety posture, see CAPABILITIES.md.
- For the documented invocations —
configure,build,modify,run,test,debug, plus theCommand appendix— see TASKS.md.
Related skills
doca-pcc— the host-side library for writing and loading custom congestion-control kernels onto the DPA. It is a SEPARATE surface: the firmware PCC diagnostic counterspcc_counters.shreads exist independently of any customdoca-pcckernel, but an operator tuning a custom algorithm may read these counters to observe device-level CC behaviour. Conflating the script (firmware counter readout) with the library (custom algorithm load/control) is the most common PCC first-touch error.doca-public-knowledge-map— routing to the public DOCA / PCC documentation set, including the firmware PCC algorithm configuration.doca-version— canonical DOCA version-handling rules. The## Version compatibilitysection inCAPABILITIES.mdis a concise overlay that redirects here.doca-setup— env preparation, install verification, mst tools, debugfs, and the I have no install yet path with the public NGC DOCA container. This skill assumes its preconditions are satisfied.doca-debug— the cross-cutting debug ladder. The PCC counter readout slots in as a read-only device-state evidence source before any congestion-control tuning recommendation is made.doca-programming-guide— general DOCA programming patterns shared across the bundle.
评论 (0)
暂无评论，成为第一个评论者吧！