Indexed by skills.sh from answerzhao/agent-skills

Extract false-positive and false-negative gaps from VLM binary-classification-question (BCQ, yes/no) predictions. Use when the user asks to "analyze VLM BCQ gaps", "extract VLM false positives and false negatives", or identify failure cases from a predictions JSON for DEFT roo…

Analyzes and understands video content using Volcengine LAS Doubao vision-language models (VLM). Multimodal AI video analysis, video comprehension, and visua...

Visual inspection helper for VLM and OCR workflows. Use when agent needs to help a vision model see an image more clearly before re-analysis: rotate misalign...

Extract false-positive and false-negative gaps from VLM binary-classification-question (BCQ, yes/no) predictions. Use when the user asks to "analyze VLM BCQ gaps", "extract VLM false positives and false negatives", or identify failure cases from a predictions JSON for DEFT roo…

Practical guidance for training MoE VLMs in Megatron Bridge. Compares FSDP and 3D-parallel approaches, using rounded lessons from Qwen3-VL, Qwen3-Next, and other multimodal experiments.

Use GLM-4.7V's multimodal grounding capability to detect and locate objects/text in images. Activate when user asks to find, locate, detect, or ground specif...

Practical guidance for training MoE VLMs in Megatron Bridge. Compares FSDP and 3D-parallel approaches, using rounded lessons from Qwen3-VL, Qwen3-Next, and other multimodal experiments.

Indexed by skills.sh from nvidia/skills

Indexed by skills.sh from nvidia/skills

Use the VLM Run CLI (`vlmrun`) to interact with Orion visual AI agent. Process images, videos, and documents with natural language. Triggers: image understanding/generation, object detection, OCR, video summarization, document extraction, image generation, visual AI chat, 'gen…

Two-step image grounding pipeline: extracts referring expressions from (image, caption) pairs and grounds them to pixel-space bounding boxes via a VLM. Use when the user wants to ground captions to bboxes, generate phrase-grounded annotations, auto-label images for grounding, …

Four-step image referring-expression pipeline: turns images plus KITTI bounding-box labels into region descriptions, scene captions, grounded referring expressions, and (optionally) verified expressions via VLM distillation. Use when the user wants to generate referring-expres…

Multi-step video annotation pipeline that turns raw videos into Chain-of-Thought training data — multi-level captions, structured descriptions, and QA pairs (MCQ, binary, open-ended) with reasoning traces, via VLM/LLM distillation. Use when the user wants to "create video trai…

OCR and text extraction from Word documents (.docx, .doc) using the MinerU API. This skill leverages mineru-open-api CLI to perform optical character recogni...

Run `tao-daft convert` to convert NVIDIA TAO DAFT datasets between supported formats. Do not use for non-DAFT data. Use when the user asks to convert a DAFT dataset, change DAFT format, change a TAO dataset format, or run `tao-daft convert`.

Cosmos-Embed1 video-text embedding for text-to-video retrieval, video-to-video search, semantic deduplication, and fine-tuning. Use when the user asks to "fine-tune Cosmos-Embed1", "run cosmos-embed inference", "export Cosmos-Embed1", "embed videos", or "search videos with text".

Cosmos3-Nano video QA supervised fine-tuning with FSDP parallelism. Use when training or evaluating video question-answering models, fine-tuning Cosmos3-Nano or compatible Cosmos Reason models with SFT/LoRA, or working with Cosmos-RL. Trigger phrases include "fine-tune Cosmos"…

This skill should be used when the user asks to "build a frame extraction job" / "视频抽帧 / 抽关键帧", "label driving images with a VLM" / "图像打标 / image labeling wi...

Extract brand identity (colors, typography, visual style, imagery) from any website URL. Scrapes the site, analyzes CSS/images with K-means and VLM, and retu...

Extract brand identity (colors, typography, visual style, imagery) from any website URL. Scrapes the site, analyzes CSS/images with K-means and VLM, and retu...

Use MiniMax Coding Plan API for real-time web search and image understanding (VLM). Based on yorch233/minimax-coding-plan-tool, patched to use api.minimax.io...

Analyze PDF files using MiniMax API. Supports text extraction, keyword search, and image-based VLM analysis (converts PDF pages to images first). Requires Mi...

Professional-grade OCR for PDFs and images using MinerU. Advanced text recognition with VLM (Vision Language Model) support for complex layouts, mixed conten...

Quantum experiment VISUAL review and Question Answering (VQA) using Vision Language Models (VLM). Analyze experiment RESULT PLOTS to: (1) Describe plot types...

Use VLM Run (vlmrun) to generate transcriptions from YouTube videos. Download a video with yt-dlp, then run vlmrun to transcribe with optional timestamps. VLMRUN_API_KEY must be in .env; follow vlmrun-cli-skill for CLI setup and options.

Pick the serving stack and per-runtime memory flags (vLLM, SGLang, llama.cpp, TensorRT Edge-LLM) for an LLM/VLM workload on any NVIDIA Jetson.

Benchmark Jetson LLM/VLM serving performance across vLLM, llama.cpp, and Ollama with structured JSON output.

Use when calling local AI on this Mac — text generation, embeddings, speech-to-text, OCR, or image understanding. LLM/VLM via oMLX gateway at localhost:8000/...

Choose the right MoE token dispatcher (`alltoall`, DeepEP, or HybridEP) for the hardware, EP degree, and optimization stage. Summarizes patterns from DSV3, Qwen3, Qwen3-Next, and VLM bring-up work.

Extract text from images, documents and scanned PDFs using OpenOCR - supports text detection, recognition, universal VLM recognition, and document parsing with layout analysis

Extract text from images, documents and scanned PDFs using OpenOCR - a lightweight and efficient OCR system with document parsing model requiring only 0.1B parameters, capable of running recognition on personal PCs. Supports text detection, recognition, universal VLM recogniti…

Multi-step video annotation pipeline that turns raw videos into Chain-of-Thought training data — multi-level captions, structured descriptions, and QA pairs (MCQ, binary, open-ended) with reasoning traces, via VLM/LLM distillation. Use when the user wants to "create video trai…

Build travel destination scenarios and brochures from a city name. Fetches street-level and landmark imagery from OpenStreetCam and Wikimedia Commons, then uses VLM Run (vlmrun) to generate a travel video and a travel plan. Use when the user wants a travel brochure, destinatio…

Use this skill when deploying standalone RT-VLM dense captioning or calling its REST API (uploads, captions, streams, chat-completions, Kafka). Not for VSS profile deploy or video-search ingestion.

Use this skill when producing a VSS analysis report — Mode A per-clip VLM, Mode B incident-range via video-analytics. Not for standalone video summarization, real-time alerts or ad-hoc Q&A.

Use this skill when reading video-analytics metrics, incidents, alerts, and sensor data via the VA-MCP server (port 9901). Not for live VLM or incident-range narrative reports.

Use to summarize a recorded video via the LVS summarization microservice (HITL-gated) with a VLM fallback. Not for report generation or live RTSP captioning.

Extracts structured weekly menu content (breakfast, lunch, snacks by day) from a cafeteria menu image using a VLM via OpenRouter. Use when the user provides...

Pick the serving stack and per-runtime memory flags (vLLM, SGLang, llama.cpp, TensorRT Edge-LLM) for an LLM/VLM workload on any NVIDIA Jetson.

Benchmark Jetson LLM/VLM serving performance across vLLM, llama.cpp, and Ollama with structured JSON output.

Choose the right MoE token dispatcher (`alltoall`, DeepEP, or HybridEP) for the hardware, EP degree, and optimization stage. Summarizes patterns from DSV3, Qwen3, Qwen3-Next, and VLM bring-up work.

NVIDIA RAG Blueprint — deploy, configure, troubleshoot, and manage. Handles any RAG action: deploy, install, start, enable, disable, toggle, change, configure, troubleshoot, debug, fix, shutdown, stop, or tear down any RAG feature or service (Agentic RAG, VLM, guardrails, quer…

Fine-tune any HuggingFace CV / VLM / LLM model on local NVIDIA GPUs inside an NGC PyTorch container. Use when the user wants to fine-tune a HuggingFace model (full or LoRA), train a vision / VLM / LLM model end-to-end, generate a reproducible HF training pipeline, smoke-test a…

Use this skill when deploying standalone RT-VLM dense captioning or calling its REST API (uploads, captions, streams, chat-completions, Kafka). Not for VSS profile deploy or video-search ingestion.

Use this skill when producing a VSS analysis report — Mode A per-clip VLM, Mode B incident-range via video-analytics. Not for standalone video summarization, real-time alerts or ad-hoc Q&A.

Use to call the VIOS REST API (sensor list, timelines, clip extraction, snapshots, add/delete sensors and streams). Not for VLM inference or search.

Use this skill when reading video-analytics metrics, incidents, alerts, and sensor data via the VA-MCP server (port 9901). Not for live VLM or incident-range narrative reports.

Use to summarize a recorded video via the LVS summarization microservice (HITL-gated) with a VLM fallback. Not for report generation or live RTSP captioning.