nvidia--cosmos · Horonomos-laps - Agent Skill

01 前言 Front Matter

name: nvidia--cosmos
description: NVIDIA Cosmos — open platform world models cho Physical AI: robotics, autonomous vehicles. Cosmos3-Nano (16B) + Super (64B). Text/video/action I/O. Diffusers/vLLM/NIM. OpenMDW-1.1.
allowed-tools: Bash, Read, Write
user-invocable: true

03 SKILL.md 正文

NVIDIA Cosmos: world model platform cho Physical AI — robots và autonomous vehicles hiểu + simulate thế giới vật lý thay vì chỉ generate text.

Models — Cosmos 3

Cosmos3-Nano  (16B)  — compact, multimodal understanding + world simulation
Cosmos3-Super (64B)  — frontier-scale, advanced capabilities

Specialized:
  Text-to-image variants
  Image-to-video variants
  Robot policy models

2 Runtime Surfaces

Reasoner
  Input:  text + vision
  Output: text
  Uses:   world understanding, physical reasoning, task planning, autonomous decisions
  → "Is this grasp stable?" / "What happens if I push this box?"

Generator
  Input:  text + vision + sound + actions
  Output: visual content + audio + action sequences
  Uses:   simulation, synthetic training data, video generation

Deployment Options

# Diffusers (research/dev)
pip install diffusers transformers
# huggingface: nvidia/Cosmos3-Nano

# vLLM-Omni (production API)
pip install vllm
vllm serve nvidia/Cosmos3-Nano --task generate

# NIM containers (turnkey)
docker run nvcr.io/nim/nvidia/cosmos3-nano:latest

# Cosmos Framework (training + eval)
git clone https://github.com/NVIDIA/cosmos
pip install -e .

Key Capabilities

Video captioning + event detection
Embodied reasoning + physical plausibility analysis
Spatial grounding
Text-to-video + image-to-video generation
Action-conditioned simulation
Synthetic training data generation for robotics

Khi nào dùng

Research: physical AI, embodied reasoning
Generate synthetic training data cho robot models
Simulate robot scenarios trước khi deploy hardware
Video generation có physics-aware

Khi KHÔNG dùng

General chatbot — dùng Claude
Text generation — dùng Claude/GPT
Requires NVIDIA GPU (A100/H100 recommended cho Super)

License

OpenMDW-1.1 (custom NVIDIA license, not fully open-source)

Source

https://github.com/NVIDIA/cosmos · OpenMDW-1.1 · +479⭐/week

04 同仓库更多 Skills

Horonomos-laps/Yana-AI

openai--sentry--sentry

Use when the user asks to inspect Sentry issues or events, summarize recent production errors, or pull basic Sentry health data via the Sentry API; perform read-only queries with the bundled script and require `SENTRY_AUTH_TOKEN`.

★ 13

Horonomos-laps/Yana-AI

ito-basket-compare

Compare Itô prediction-market baskets against a user's knowledge base, portfolio notes, financial context, watchlist, or research thesis. Use for read-only basket comparison and gap analysis without investment advice or live trading.

★ 13

Horonomos-laps/Yana-AI

implementing-network-traffic-baselining

Build network traffic baselines from NetFlow/IPFIX data using Python pandas for statistical analysis, z-score anomaly detection, and hourly/daily traffic pattern profiling

★ 13

Horonomos-laps/Yana-AI

detecting-evasion-techniques-in-endpoint-logs

Detects defense evasion techniques used by adversaries in endpoint logs including log tampering, timestomping, process injection, and security tool disabling. Use when investigating suspicious endpoint behavior, building detection rules for evasion tactics, or conducting threat hunting for stealthy adversary activity. Activates for requests involving evasion detection, defense evasion analysis, log tampering detection, or MITRE ATT&CK TA0005.

★ 13

Horonomos-laps/Yana-AI

implementing-microsegmentation-with-guardicore

Implementing microsegmentation using Akamai Guardicore Segmentation to map application dependencies, create granular network policies, visualize east-west traffic flows, and enforce least-privilege communication between workloads across data centers and cloud.

★ 13

Horonomos-laps/Yana-AI

ramdisk-tmpfs-patterns

暂无简介

★ 13