NVIDIA Nemotron is a developer resource for building with the Nemotron family of AI models, providing training recipes, deployment guides, datasets, cookbooks, and end-to-end examples. It is intended for developers and researchers training, customizing, deploying, or applying Nemotron models to agentic AI use cases. Its catalogue add-ons include a Claude Code plugin that guides users through Nemotron customization steps.
Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/NVIDIA-NeMo/Nemotronnpx agentmods add skills/nvidia-nemo/nemotron/nemotron-add-patternWrote 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-nemo/nemotron/nemotron-add-pattern)<a href="https://agentmods.dev/skills/nvidia-nemo/nemotron/nemotron-add-pattern"><img src="https://agentmods.dev/badge/skills/nvidia-nemo/nemotron/nemotron-add-pattern/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/nvidia-nemo/nemotron/nemotron-add-pattern"><img src="https://agentmods.dev/badge/skills/nvidia-nemo/nemotron/nemotron-add-pattern.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00058 | $0.01399 |
| Opus 5 | $0.00029 | $0.00700 |
| Sonnet 5 | $0.00012 | $0.00280 |
| Haiku 4.5 | $0.00006 | $0.00140 |
Grade A, and why
nemotron-add-pattern 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 9d 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 — 153 lines — stays where its author put it; the contents beside it link to each section on GitHub.
nemotron-add-pattern
Invocation: /nemotron-add-pattern.
You help contributors add a new cross-cutting pattern to src/nemotron/steps/patterns/ without getting the frontmatter, scope, catalog regeneration, or tests wrong.
Tone
Concise. Checklist-first. Ask for missing facts before writing files.
- Status updates: ≤2 lines
- Prefer bullets over long prose
- Say exactly which pattern file you will create and which commands you will run
- Do not guess step ids or confidence level
- Keep the recommendation actionable, not academic
- Always regenerate
PATTERNS.mdand run tests
Workflow
Four phases. Always in this order.
1. Orient
Read these first:
src/nemotron/steps/patterns/sft-small-dataset-prefer-lora.mdsrc/nemotron/steps/PATTERNS.mdsrc/nemotron/steps/index.pytests/steps/test_patterns.py
Then ask the contributor:
- What is the pattern about? (one sentence)
- When should it apply? (natural-language triggers)
- Which steps does it touch? (step ids, or
[]for global) - What is the confidence level? (
high,medium, orexperimental) - Does it introduce a new concept or just encode existing tribal knowledge?
Use these repo conventions:
- Pattern files live at
src/nemotron/steps/patterns/{id}.md. - The filename stem must match the frontmatter
id. - Required frontmatter fields are
id,title,tags,triggers,steps, andconfidence. steps: []is valid for a global pattern.- Valid confidence values are
high,medium, andexperimental. - The body uses these sections:
## When to apply,## What to do,## Exceptions,## References. - Step-strategy cross-links in
step.tomlare a separate task. Do not edit them here.
2. Generate
Create:
src/nemotron/steps/patterns/{id}.md
The pattern file must contain:
- YAML frontmatter with
id,title,tags,triggers,steps,confidence ## When to apply## What to do## Exceptions## References
Generation rules:
- Keep the pattern id kebab-case and make it match the filename exactly.
- Turn vague triggers into 2–4 concrete, observable conditions.
- Scope the pattern honestly: use explicit step ids if it only applies to a subset of steps; use
[]only when it is truly global. - Put the recommendation itself in
What to do; keep background explanation shorter than the action guidance. - If the pattern introduces a new concept, define it in the first paragraph of
When to apply. - Do not modify existing patterns.
- Regenerate the catalog with:
uv run python src/nemotron/steps/index.py
- Run validations with:
uv run pytest tests/steps -q
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.
- 9d ago First seen · 153 lines · 58 tokens per session scan A f9b39404a40a
nemotron-add-pattern is a skill published in the GitHub repository NVIDIA-NeMo/Nemotron (2,032 stars, last pushed 2d ago), licensed Apache-2.0. It adds 58 tokens to every session and 1,399 once invoked, about $0.0003 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 2026-08-30.
Other skills, from other repositories
tinker-fine-tuning
Provides guidance for fine-tuning LLMs using the Tinker cloud training API from Thinking Machines Lab. Use when running supervised fine-tuning, reinforcement learning (GRPO/PPO), or LoRA training on cloud GPUs via Tinker's managed infrastructure instead of local compute.
fine-tuning-with-trl
Fine-tune LLMs using reinforcement learning with TRL - SFT for instruction tuning, DPO for preference alignment, PPO/GRPO for reward optimization, and reward model training. Use when need RLHF, align model with preferences, or train from human feedback. Works with HuggingFace Transformers.
fine-tuning-with-trl
Fine-tune LLMs using reinforcement learning with TRL - SFT for instruction tuning, DPO for preference alignment, PPO/GRPO for reward optimization, and reward model training. Use when need RLHF, align model with preferences, or train from human feedback. Works with HuggingFace Transformers.
fine-tuning-with-trl
Fine-tune LLMs using reinforcement learning with TRL - SFT for instruction tuning, DPO for preference alignment, PPO/GRPO for reward optimization, and reward model training. Use when need RLHF, align model with preferences, or train from human feedback. Works with HuggingFace Transformers.
fine-tuning-with-trl
Fine-tune LLMs using reinforcement learning with TRL - SFT for instruction tuning, DPO for preference alignment, PPO/GRPO for reward optimization, and reward model training. Use when need RLHF, align model with preferences, or train from human feedback. Works with HuggingFace Transformers.
model_finetuning
Fine-tune LLMs using reinforcement learning with TRL - SFT for instruction tuning, DPO for preference alignment, PPO/GRPO for reward optimization, and reward model training. Use when need RLHF, align model with preferences, or train from human feedback. Works with HuggingFace Transformers.