video-analyzer

Analyze video content via frames or streams using vision-language models.

11|4|Updated Mar 3, 2026
One-click install
npx skills add https://github.com/zrong/skills --skill video-analyzer-zrong
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: video-analyzer
Source: https://github.com/zrong/skills/tree/main/video-analyzer
Command: npx skills add https://github.com/zrong/skills --skill video-analyzer-zrong

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires openai, opencv-python-headless, httpx, yt-dlp, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill solves the difficulty of manually reviewing, summarizing, or extracting specific information from long or complex video files by leveraging advanced vision-language models.

Core Features & Use Cases

  • Multi-Source Analysis: Processes local video files, direct URLs, and streaming platform links (YouTube, Bilibili) seamlessly.
  • Flexible Analysis Modes: Supports both frame-based sampling for cost-efficiency and direct video input for high-fidelity understanding.
  • Structured Output: Capable of returning JSON-formatted analysis, making it ideal for automated workflows like identifying animation loops or extracting metadata.

Quick Start

Use the video-analyzer skill to analyze the provided video file and summarize its key events in a structured format.

Frequently Asked Questions about video-analyzer

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

FAQPage Schema
How do I analyze video content from a YouTube or Bilibili URL?

You can analyze video content from streaming platform URLs by providing the direct link. The tool uses yt-dlp to handle external video source downloads and extracts frames for vision-language model processing.

Can I extract structured JSON data from a video file?

Yes, you can extract structured JSON data from a video file. The analysis supports returning JSON-formatted output, making it ideal for automated workflows like identifying animation loops or extracting metadata.

How does frame-based video analysis work with vision-language models?

Frame-based video analysis works by using OpenCV for frame manipulation to extract specific frames from the video. These frames are then processed through multimodal vision-language models to provide summaries and Q&A.

Do I need OpenCV and yt-dlp to process local video files?

You need OpenCV for frame manipulation and yt-dlp for handling external video source downloads. While local files do not require yt-dlp for downloading, OpenCV is still required for extracting frames from the local video stream.

What is the best way to summarize key events in a long video?

The best way to summarize key events is to use frame-based sampling for cost-efficiency. This approach extracts frames at intervals and passes them through a vision-language model to generate a structured summary.

Does video analysis support direct video stream input for high fidelity?

Yes, video analysis supports direct video input for high-fidelity understanding. You can process raw video streams through vision-language models instead of using frame-based sampling to achieve more detailed results.