NVIDIA/skills is a catalogue of portable instruction sets that teach coding agents how to use NVIDIA software for robotics, simulation, CUDA, retrieval-augmented generation, and related workflows. Developers install these skills in agents such as Claude Code or Codex, while the catalogue mirrors skills maintained in separate NVIDIA product repositories.
Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/nvidia/skills/hsb-ip-packetizernpx skills add NVIDIA/skills --skill hsb-ip-packetizergit clone --depth 1 https://github.com/NVIDIA/skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/nvidia/skills/hsb-ip-packetizer)<a href="https://agentmods.dev/skills/nvidia/skills/hsb-ip-packetizer"><img src="https://agentmods.dev/badge/skills/nvidia/skills/hsb-ip-packetizer.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00040 | $0.03484 |
| Opus 5 | $0.00020 | $0.01742 |
| Sonnet 5 | $0.00008 | $0.00697 |
| Haiku 4.5 | $0.00004 | $0.00348 |
Grade A, and why
hsb-ip-packetizer scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 229 lines — stays where its author put it; the contents beside it link to each section on GitHub.
HSB IP Packetizer Skill
Purpose
Choose packetizer-related HOLOLINK_def.svh fields for the NVIDIA Holoscan Sensor Bridge Sensor RX path.
This skill owns only the packetizer slice of HOLOLINK_def.svh:
SIF_RX_PACKETIZER_EN[]SIF_RX_VP_COUNT[]SIF_RX_SORT_RESOLUTION[]SIF_RX_VP_SIZE[]SIF_RX_NUM_CYCLES[]
Use hsb-ip-def for whole-file generation, validation, comparison, and non-packetizer macros.
Prerequisites
SENSOR_RX_IF_INSTmust be defined and nonzero before packetizer fields are relevant.- Per-RX-interface
SIF_RX_WIDTH[]values and the user's data manipulation intent are required before deriving enabled packetizer values. - Live packetizer RTL is preferred for source-sensitive claims; bundled references cover known rev
16'h2604and compatible rev16'h2603. - Whole-file generation and validation remain the responsibility of
hsb-ip-def.
Instructions
- Only configure packetizer fields when
SENSOR_RX_IF_INSTis defined. If Sensor RX is disabled, say no packetizer fields are needed. - First classify the request as Standalone SVH, Def-Skill Handoff, Explain, or runtime APB pattern-RAM. For Def-Skill Handoff, load
references/handoff-contract.mdand makepacketizer_profile_overlaythe first non-empty output after any required clarification. - Ask one question per turn only while collecting the data-description facts needed to derive a grounded profile. Prefer deriving macro values from those facts instead of requesting per-macro confirmation.
- Use sensor-agnostic language. Say "sensor data", "data word", "stream", "bandwidth", and "packetization"; use camera-specific language only after the user says the sensor data is image/camera data.
- Never silently default enabled packetizer fields. Derive them from the user's stated data layout and manipulation intent, print the complete field set in one block, and then explain what the selected settings enable and what assumptions they encode.
- Avoid ungrounded "typical", "common", "most designs", or corpus-frequency claims. Anchor statements to HSB docs, RTL behavior, or the user's stated requirements.
- Keep runtime APB pattern-RAM programming out of v1 generation.
HOLOLINK_def.svhenables/sizes hardware; runtime software still controls pattern RAM, virtual-port selection, sort controls, bypass, padding, duplication, and latency. - In a handoff from
hsb-ip-def, use facts already known by that def flow. If the prompt gives RX count, enabled/pass-through intent, lane or stream width, virtual stream count, and cycle window, emit the overlay without re-asking for macro names or whole-file details. Leave final merge and full validation tohsb-ip-def.
What ships with it
7 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 229 lines · 40 tokens per session scan A c1ce9142b917
hsb-ip-packetizer is a skill published in the GitHub repository NVIDIA/skills (3,206 stars, last pushed today), licensed Apache-2.0. It adds 40 tokens to every session and 3,484 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.
Other skills, from other repositories
spark-training-gotchas
Preflight and diagnose the ten known failure modes for ML training on NVIDIA DGX Spark. Use when a training run on DGX Spark fails to start, OOMs below the 128GB limit, slows down mid-run, or before any multi-hour training job on GB10.
terraform-module-library
Build reusable Terraform modules for AWS, Azure, GCP, and OCI infrastructure following infrastructure-as-code best practices. Use when creating infrastructure modules, standardizing cloud provisioning, or implementing reusable IaC components.
readable-verilog-generator
Use when creating, writing, reviewing, annotating, repairing, refactoring, or validating readable Verilog RTL, including synthesizable Verilog-2001 .v files, existing-RTL analysis, semantic comment annotation, testbench scaffold planning, ASIC-quality review, local or remote Vivado/xsim validation, evidence-backed…
technical-troubleshooting
Provide setup, troubleshooting, and maintenance guidance. Use when the user reports a device that won't power on, connectivity issues, setup questions, overheating, or maintenance concerns.
gpu-optimizer
GPU optimization for consumer NVIDIA GPUs (8-24GB VRAM) covering mixed precision, gradient checkpointing, XGBoost GPU, CuPy/cuDF migration, and torch.compile. Triggers on: "optimize GPU training", "speed up CUDA", "reduce OOM", "migrate NumPy to CuPy", "manage GPU memory", "benchmark PyTorch".
mdm-device-management
Manage and secure company devices with MDM solutions — enroll macOS, Windows, iOS, and Android devices, enforce security policies, and automate software deployment. Use when setting up device management for a growing team.