local-video-inspection

Inspect local video files by sampling frames and extracting transcripts without uploading media.

Updated Sep 6, 2026
One-click install
npx skills add https://github.com/rickvang/Persona-Library --skill local-video-inspection-rickvang
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: local-video-inspection
Source: https://github.com/rickvang/Persona-Library/tree/main/.agents/skills/local-video-inspection
Command: npx skills add https://github.com/rickvang/Persona-Library --skill local-video-inspection-rickvang

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve? When you cannot play a video or need to know what a local video contains, this Skill summarizes it from representative frames and available transcripts while keeping the file entirely on your machine. ## Core Features & Use Cases - Frame Sampling: Extracts 8-12 uniformly distributed interior frames using FFmpeg or a bundled Windows Media Foundation PowerShell extractor, with targeted timestamps for unclear sections. - Transcript Resolution: Reads sidecar .srt/.vtt/.txt/.md files first, then embedded subtitle streams via FFmpeg, and optionally local Whisper speech-to-text using only pre-installed models. - Use Case: You receive a screen recording on a Windows machine without FFmpeg and playback is unavailable; the Skill samples frames via Media Foundation, finds a matching .srt sidecar, and returns a summary with transcript provenance and confidence. ## Quick Start Use the local-video-inspection skill to summarize what the video at C:\Videos\demo.mp4 contains from sampled frames and any available transcript.

Frequently Asked Questions about local-video-inspection

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

FAQPage Schema
How do I summarize a local video without uploading it?▼

Sample 8-12 uniformly distributed frames with a local decoder, review them with an image-inspection tool, and resolve transcript evidence from sidecar files or embedded subtitles. The summary labels frame evidence, transcript evidence, and synthesis separately with confidence and limitations.

How to extract frames from video on Windows without FFmpeg?▼

Run the bundled extract-video-frames.ps1 script with Windows PowerShell 5.1, which uses the Windows Media Foundation API to generate thumbnails. It emits JSON with duration, dimensions, sampling mode, and frame paths, and supports targeted timestamps via -TimestampsSeconds.

Does the video transcript extraction support Whisper speech-to-text?▼

Yes, but only when you pass -AllowSpeechToText and an OpenAI Whisper-compatible CLI, FFmpeg, and an existing local model file are already installed. The helper returns unavailable rather than downloading a missing model or dependency.

Can this skill edit or convert video files?▼

No, it is strictly for inspection and summarization, not video editing or remote-video research. It never modifies, renames, moves, deletes, or overwrites the source video, and writes frames only to an explicit output directory.

Why does frame extraction fail in PowerShell Core?▼

The extractor requires Windows PowerShell 5.1 Desktop edition because it relies on WinRT Media Foundation interop. Invoke C:\Windows\System32\WindowsPowerShell\v1.0\powershell.exe explicitly, or use an installed FFmpeg pair instead.