Where are the processed video and audio files stored?

By default, processed outputs are stored in '/tmp/processed_videos'. You can customize this destination by setting the 'VIDEO_OUTPUT_DIR' environment variable.

How does the DuckDB integration work?

The skill automatically records every processing event into a DuckDB table named 'fs_events', storing the file path, event type, file size, checksum, and timestamp for easy querying and auditing.

Can I use this for AI multimodal analysis?

Yes, it is designed for this purpose. By extracting thumbnails and audio tracks automatically, it prepares the necessary assets for Claude to perform multimodal analysis on video content.

What dependencies are required for the Video Processor skill?

You must have Babashka, FFmpeg (including ffprobe), and DuckDB installed on your system. These can be easily installed via package managers like Homebrew using 'brew install babashka ffmpeg duckdb'.

Does this skill support batch processing?

Yes, the skill can be invoked on a single video file or configured to watch an entire directory to automatically process any new media files as they are added.

Video Processing Pipeline

Name: Video Processing Pipeline
Author: plurigrid

byplurigrid

•

데이터 과학 및 ML

Automates video metadata extraction, thumbnail generation, web-optimized transcoding, and audio extraction with integrated DuckDB tracking.

소개

The Video Processor skill provides a robust, scriptable pipeline for managing video assets using Babashka, FFmpeg, and DuckDB. It streamlines the preparation of video content for web playback and multimodal AI analysis by automatically generating thumbnails, transcoding to H.264, and extracting audio tracks. It maintains a searchable historical record of processing events, checksums, and metadata within a local DuckDB instance, making it ideal for building automated ingestion systems and searchable media indexes.

주요 기능

Real-time directory watching capabilities for automated media ingestion workflows
Automated thumbnail generation at the 5-second mark and high-quality MP3 audio extraction
Automated metadata extraction via ffprobe including codec, bitrate, and resolution
2 GitHub stars
Fast-start web-optimized H.264/AAC transcoding for streaming compatibility
Integrated DuckDB database for persistent tracking of processing events and session history

사용 사례

Building a searchable local database of video assets and technical metadata for large media libraries
Extracting audio tracks and frame samples for multimodal LLM analysis and training
Preparing raw video footage for web-based streaming and cross-platform playback

소개

주요 기능

Real-time directory watching capabilities for automated media ingestion workflows
Automated thumbnail generation at the 5-second mark and high-quality MP3 audio extraction
Automated metadata extraction via ffprobe including codec, bitrate, and resolution
2 GitHub stars
Fast-start web-optimized H.264/AAC transcoding for streaming compatibility
Integrated DuckDB database for persistent tracking of processing events and session history

사용 사례

Building a searchable local database of video assets and technical metadata for large media libraries
Extracting audio tracks and frame samples for multimodal LLM analysis and training
Preparing raw video footage for web-based streaming and cross-platform playback