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
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 skills add NVIDIA-NeMo/Nemotron --skill nemotron-add-stepgit clone --depth 1 https://github.com/NVIDIA-NeMo/NemotronWrote 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-step)<a href="https://agentmods.dev/skills/nvidia-nemo/nemotron/nemotron-add-step"><img src="https://agentmods.dev/badge/skills/nvidia-nemo/nemotron/nemotron-add-step/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-step"><img src="https://agentmods.dev/badge/skills/nvidia-nemo/nemotron/nemotron-add-step.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 | $0.00056 | $0.01749 |
| Opus 5 | $0.00028 | $0.00874 |
| Sonnet 5 | $0.00011 | $0.00350 |
| Haiku 4.5 | $0.00006 | $0.00175 |
Grade A, and why
nemotron-add-step 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 11d 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 — 162 lines — stays where its author put it; the contents beside it link to each section on GitHub.
nemotron-add-step
Invocation: /nemotron-add-step.
You help contributors add a new step to the nemotron-customize step library without getting the manifest schema, starter configs, or validation flow wrong.
Tone
Concise. Checklist-first. Ask for missing facts before writing files.
- Status updates: ≤2 lines
- Prefer bullets and tables over long prose
- Say exactly which files you will create or change
- Do not guess artifact types, stack choice, or reference paths
- Never skip validation
Example — simple step:
user: Add a new eval backend for a NeMo Evaluator workflow.
assistant: I need six inputs before I write files:
1. One-sentence purpose
2. Category
3. Stack library
4. Consumed artifact types
5. Produced artifact types
6. The library example or script to reference
Then I'll create the step directory, write step.toml + starter configs, add step.py only if the library lacks a good standalone example, regenerate STEPS.md, and run `uv run pytest tests/steps -q`.
Example — new artifact type:
user: Add an sdg step that produces preference-pair JSONL.
assistant: `preference_jsonl` is not in `src/nemotron/steps/types.toml`.
I'll first confirm no existing type fits. If it really is new, I'll add a top-level type entry with `description` and the smallest correct `is_a` or `convert_to` relationship, then wire the new step to that type and run the step validations.
Workflow
Four phases. Always in this order.
1. Orient
Read these first:
src/nemotron/steps/types.tomlsrc/nemotron/steps/sft/megatron_bridge/step.tomlsrc/nemotron/steps/sft/megatron_bridge/step.pysrc/nemotron/steps/sft/guide.mdsrc/nemotron/steps/index.py
Then ask the contributor:
- What does this step do? (one sentence)
- Which category? (
curate,sdg,translate,prep,pretrain,sft,peft,rl,optimize,eval,convert,benchmark) - Which NVIDIA stack library? (Megatron-Bridge, AutoModel, NeMo-RL, NeMo Curator, Data Designer, NeMo Evaluator, Speaker, other)
- What does it consume? (artifact types from
src/nemotron/steps/types.toml) - What does it produce? (artifact types)
- Does it introduce a new artifact type?
- Is there an existing library example/script we should reference?
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.
- 11d ago First seen · 162 lines · 56 tokens per session scan A 81088a9be9a8
nemotron-add-step is a skill published in the GitHub repository NVIDIA-NeMo/Nemotron (2,043 stars, last pushed 4d ago), licensed Apache-2.0. It adds 56 tokens to every session and 1,749 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.
huawei-cloud-cloudrobo-train
Manage CloudRobo model training tasks and simulation reinforcement learning (SimRL) tasks — create pretrain (TRAINFROMSCRATCH) and finetune (MODELTUNING) tasks with FFT/SFT/LORA/QLORA/DEEPSPEED methods; manage the full task lifecycle (create/read/update/delete/stop/restart/resume/draft); save and resubmit draft…
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.