ocr-toolkit

Agent Skills for visual AI tasks including image understanding, video processing, document extraction, and multi-modal generation using VLM Run's Orion agent

Turn videos into a sequence of relevant still frames + transcript + a self-contained HTML report so Claude can view them as images, hear the audio, and write its analysis back into the report. Pass a local path, an http(s) URL, or pipe video bytes on stdin.

Image and visual analysis with screenshot interpretation and text extraction

Let Claude watch a video from YouTube, Instagram, X, Vimeo, TikTok or any yt-dlp site. Downloads with yt-dlp, extracts auto-scaled frames with ffmpeg, transcribes locally with mlx-whisper (no API key, nothing leaves your machine), and hands frames + transcript to Claude.

Compose yt-dlp + ffmpeg + Whisper into a single command that hands an AI agent the raw materials to watch any social video — VIDEO + FRAMES + TRANSCRIPT, ready for an LLM to read frames as images and transcript as text.

Give Claude the ability to watch and understand videos — extracts frames and audio for full video perception