skills.rest
NVIDIA-AI-Blueprints
NVIDIA-AI-BlueprintsOFFICIAL
Explore all skills by this author →
WebsiteGitHub

vss-generate-video-report-rag

Generates VSS video summary reports with LVS HITL and Enterprise RAG document grounding.

1.8k|384|Updated Oct 22, 2024
Search Tags:#report generation#docker compose#video summarization#video analytics#enterprise rag#hitl workflow#nvidia vss
One-click install
npx skills add https://github.com/NVIDIA-AI-Blueprints/video-search-and-summarization --skill vss-generate-video-report-rag
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: vss-generate-video-report-rag
Source: https://github.com/NVIDIA-AI-Blueprints/video-search-and-summarization/tree/main/skills/vss-generate-video-report-rag
Command: npx skills add https://github.com/NVIDIA-AI-Blueprints/video-search-and-summarization --skill vss-generate-video-report-rag

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Generating knowledge-enhanced video summary reports from the NVIDIA VSS blueprint requires deploying the RAG-enabled LVS profile, configuring Enterprise RAG endpoints, and driving a multi-step human-in-the-loop (HITL) workflow over HTTP. This Skill automates that entire process so users can produce document-grounded video analysis reports without memorizing API calls or config paths.

Core Features & Use Cases

  • RAG-enabled LVS deployment: Configures and deploys the LVS developer profile with the shipped config_rag.yml, including RAG_SERVER_URL and KNOWLEDGE_COLLECTION settings, while keeping secrets out of generated.env and resolved.yml.
  • Guided HITL report workflow: Walks through the full HTTP HITL sequence—video selection, scenario, events, objects of interest, and confirmation—using /v1/chat and execution polling endpoints.
  • Enterprise RAG grounding: Enriches video reports with SOP, policy, or procedure context via the frag_retrieval capability of the RAG-enabled agent config.
  • Use Case: A warehouse safety team asks for a report on forklift incidents in a recorded video, grounded in their internal safety SOPs stored in an Enterprise RAG collection; the Skill deploys the stack, collects parameters, and returns a timestamped incident report with a PDF link.

Quick Start

Ask the agent to generate a RAG-enhanced video report for a chosen video, providing the scenario, events to detect, and optional objects of interest.

Frequently Asked Questions about vss-generate-video-report-rag

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

FAQPage Schema
How do I generate a video summary report with Enterprise RAG in NVIDIA VSS?▼

Deploy the LVS developer profile with the shipped config_rag.yml, then POST to the VSS agent /v1/chat endpoint requesting a long video summarization report. Respond to the HITL prompts for scenario, events, and objects of interest, then poll the execution endpoint until the report completes.

What configuration enables Enterprise RAG document grounding for LVS reports?▼

Use the RAG-enabled agent config at deploy/docker/developer-profiles/dev-profile-lvs/vss-agent/configs/config_rag.yml. Set VSS_AGENT_CONFIG_FILE, RAG_SERVER_URL, and KNOWLEDGE_COLLECTION in generated.env; the frag_retrieval capability then grounds reports in your RAG collection.

How should I handle the RAG API key when deploying the VSS LVS profile?▼

Never write RAG_API_KEY to generated.env or resolved.yml, since Compose expands environment variables into those files. Supply it through an untracked override compose file with an ephemeral shell variable at docker compose up time, or use a secret manager.

How long does VSS long video summarization take to complete?▼

LVS report generation typically takes 3-5 minutes for a roughly 3.5-minute video. After confirming the final HITL prompt, poll the execution status endpoint every 30 seconds until the status changes from running to completed.

When should I not use the RAG report workflow for video analysis?▼

Skip this workflow for simple video understanding questions, direct LVS summarization without HITL, deployment-only tasks, or real-time alerting. Those cases are handled by the video-understanding, video-summarization, deploy, and alerts skills respectively.

Related Skills

Explore More Software Engineering Skills→
NVIDIANVIDIA
3.2k

vss-deploy-profile

Deploy, verify, debug, and tear down NVIDIA VSS profiles via Docker Compose.

Official
Advanced
kayroalexandrekayroalexandre

engineering-incident-response-commander

Coordinate engineering incident response with severity classification and post-mortem analysis.

Community
Advanced
rblake2320rblake2320

vss-ask-video

Route visual questions about video clips to the VSS video_understanding tool.

Community
Intermediate
sayalinvidiasayalinvidia

vss-summarize-video

Summarize video clips into structured summaries with events via LVS.

Community
Intermediate