vidiq_skill

Overview

This skill provides AI-powered video intelligence to download, analyze, and manipulate any video from a URL. It supports YouTube, TikTok, Instagram, X (Twitter), direct video links and local files, combining yt-dlp and ffmpeg for robust media handling. Use it to extract frames, create clips or GIFs, detect scene changes, and run vision-model analysis for content insights. Downloaded media and intermediate outputs are cached for fast reuse.

How this skill works

The tool downloads the source video (or reads a local file) and caches it to avoid repeated downloads. It uses ffmpeg to extract frames, audio, clips, GIFs, mosaics, and to detect scene changes. Extracted frames can be fed into a vision model to generate descriptions, answer questions about visual content, or produce summaries. Commands are script-driven so you can automate batching and integrate the outputs into downstream analysis pipelines.

When to use it

• When you need quick visual summaries or thumbnails from a remote video URL.
• To clip specific segments or produce short GIFs for previews or social sharing.
• When performing automated content analysis with a vision model for moderation or indexing.
• To extract audio for transcription or podcasting workflows.
• When archiving or backing up videos from multiple social platforms for research.

Best practices

• Extract more frames for long videos to get better coverage for AI analysis.
• Use scene detection with a threshold tuned to your content (e.g., 0.3) to identify logical cuts.
• Cache and reuse downloaded video files to save bandwidth and speed up repeated analyses.
• Generate GIF palettes via ffmpeg for optimized small-file animated previews.
• Keep timestamps in HH:MM:SS format when specifying frames or clips for consistency.

Example use cases

• Create a 30-second highlight clip from a long live stream by specifying start and end timestamps.
• Extract 10 evenly spaced frames from a product demo and run a vision model to produce descriptive captions.
• Detect scene boundaries in a vlog to automate chapter markers for publishing.
• Produce a compact GIF preview of a key moment for social sharing while keeping file size small.
• Pull audio from interviews as MP3 for transcription and republishing as a podcast episode.