← ClaudeAtlas

video-analyzerlisted

Analyze a video file and produce a comprehensive AI-readable report (Markdown + JSON). Use this skill whenever the user uploads or references a video file (.mp4, .mov, .avi, .mkv, .webm, .m4v) and wants to: understand its content, create a report from it, extract what is said or shown, describe what happens, count or identify characters, summarize the video for use in a Claude project (where direct MP4 upload is not supported), or generate any written analysis of video content. Trigger even for casual phrasing like "what's in this video", "read this mp4", "make a report from this video", "transcribe this video", "describe this clip", "how many characters are in this video", "who appears in this video".
VjAlbert/video-analyzer-repo · ★ 0 · Data & Documents · score 75
Install: claude install-skill VjAlbert/video-analyzer-repo
# Video Analyzer Skill Converts a video file into a complete, structured report (Markdown **and** JSON) that captures everything an AI or human needs to understand the video without watching it. Includes automatic character detection, audio transcription (Whisper), and visual analysis of key frames. --- ## Step 0 — Find the video file ```bash ls /mnt/user-data/uploads/ ``` If no video is present, ask the user to upload via the 📎 icon. --- ## Step 1 — Install dependencies & run pre-processing ```bash pip3 install openai-whisper opencv-python-headless numpy Pillow \ --break-system-packages -q 2>/dev/null || true python3 /home/claude/video-analyzer/scripts/process_video.py \ "<VIDEO_PATH>" \ --fps 0.5 \ --max-frames 30 \ --output-dir /tmp/video_work \ --model base ``` **Parameter guide:** | Parameter | Default | When to change | |-----------|---------|----------------| | `--fps` | 0.5 | Increase for fast-cut videos (try 1.0); decrease for slow/static (try 0.2) | | `--max-frames` | 30 | Increase for long videos (up to 60) | | `--model` | base | Use `small` or `medium` for better transcription accuracy | --- ## Step 2 — Run character detection ```bash python3 /home/claude/video-analyzer/scripts/detect_characters.py \ --frames-dir /tmp/video_work/frames \ --output /tmp/video_work/characters.json \ --manifest /tmp/video_work/manifest.json ``` This clusters all frames by visual similarity (HSV histogram) and outputs `characters.json