skills.rest
NVIDIA-NeMo
NVIDIA-NeMoOFFICIAL
Explore all skills by this author →
WebsiteGitHub

nemotron-add-model

Create standardized Nemotron model knowledge bases with YAML frontmatter and validated marketplace registration.

1.8k|373|Updated Oct 3, 2025
Search Tags:#knowledge base#nemotron#metadata generation#context pack#model onboarding#GitHub skills#paper chunking
One-click install
npx skills add https://github.com/NVIDIA-NeMo/Nemotron --skill nemotron-add-model
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: nemotron-add-model
Source: https://github.com/NVIDIA-NeMo/Nemotron/tree/main/skills/nemotron-add-model
Command: npx skills add https://github.com/NVIDIA-NeMo/Nemotron --skill nemotron-add-model

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It prevents contributors from adding a new Nemotron model family into the plugin ecosystem with incorrect or inconsistent knowledge artifacts (paper chunks, recipe summaries, context packs, and model card layout).

Core Features & Use Cases

  • Model-family knowledge-base onboarding: Creates the complete skills/nemotron-{model}/ shell (SKILL, INDEX, model card, paper chunks, context pack, and marketplace entry) in the same structure as existing model skills.
  • Report-faithful chunking and sourcing: Produces question-oriented paper chunks based on available HTML tech reports (falling back to PDF only when necessary).
  • Repo-convention validation and guardrails: Uses a four-phase workflow (Orient → Generate → Validate → Summarize) that enforces required frontmatter fields, cross-file indexing, and consistent currency semantics.
  • Compatibility with downstream skills: Ensures the new model family can be routed by skills like /nemotron-customize via correct context mapping and repository-native intent coverage.

Quick Start

Ask the contributor questions about the new model family slug, preferred tech-report HTML URL, architecture type, available sizes, and which existing steps should later gain [[models]] entries, then generate and validate the full skills/nemotron-{model}/ knowledge-base structure accordingly.

Frequently Asked Questions about nemotron-add-model

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I onboard a new model family knowledge base with consistent metadata and paper chunking?▼

To onboard a new model family knowledge base, you generate a standardized skills directory shell including paper chunks, context packs, and a model card. This ensures consistent indexing, YAML frontmatter validation, and correct downstream routing.

What is report-faithful chunking for tech reports and when is it needed?▼

Report-faithful chunking produces question-oriented paper chunks based on available HTML tech reports, falling back to PDF only when necessary. It is needed when adding new model knowledge bases to maintain accurate sourcing and retrieval.

How do I create a model card and context pack for a new Nemotron model family?▼

You create a model card and context pack by generating the complete nemotron model skills directory structure. This process enforces required frontmatter fields and cross-file indexing to guarantee marketplace discovery and compatibility.

Does the model onboarding workflow require YAML frontmatter for marketplace registration?▼

Yes, the model onboarding workflow requires YAML frontmatter with name and description fields for validated marketplace registration. This ensures correct currency semantics and cross-file indexing for the new model family.

Why does inconsistent knowledge artifact layout cause downstream routing failures?▼

Inconsistent knowledge artifact layout causes downstream routing failures because skills like nemotron-customize rely on correct context mapping and repository-native intent coverage. Standardized onboarding prevents these retrieval errors.

What's the best way to structure a new model knowledge base for the Nemotron plugin ecosystem?▼

The best way to structure a new model knowledge base is using a four-phase workflow that generates and validates the complete skills directory shell. This enforces repo conventions, consistent indexing, and compatibility with downstream retrieval skills.

Related Skills

Explore More Product & Management Skills→
patsnappatsnap
33

patent-family-analyzer

Analyze patent family structures and technical evolution from a patent number or URL.

Official
Advanced
FerroxLabsFerroxLabs
21

ferrox-ns-context

Route codebase intelligence requests to the correct Ferrox workflow.

Community
Intermediate
Lhhiep-maxcodeLhhiep-maxcode

nemotron-add-model

Create a Nemotron model-family knowledge base with paper chunks, recipes, and marketplace registration.

Community
Advanced
kevinsweetkevinsweet

rag-chatbot

Build RAG chatbots that ground answers in data using Elasticsearch retrieval.

Community
Advanced