What problem does it solve?
Analyzing video content with AI requires breaking recordings into discrete, reviewable pieces. This Skill converts GIF, MP4, and MOV files into timestamped PNG frames paired with aligned audio segments, plus a JSON manifest, so another agent can perform visual review or transcription without manual video scrubbing.
Core Features & Use Cases
- Interval-Based Frame Extraction: Pulls one PNG frame every N seconds (configurable, default 1s) from any ffmpeg-supported video format.
- Aligned Audio Segments: Generates per-frame AAC audio clips covering the same time window as each frame, plus a full continuous audio track for transcription.
- Structured Manifest Output: Produces manifest.json with timestamps, frame paths, and audio paths ready for handoff to a reviewing agent.
- Use Case: You have a screen recording of a user session and want an agent to review what happened. Run the extraction script, then pass the output directory to a reviewing agent that reads the manifest and analyzes each frame with its paired audio.
Quick Start
Ask the AI to extract frames and audio from your video file, for example: "Extract frames every 2 seconds from recording.mp4 into ./review-frames using the extract-video-frames skill."