← ClaudeAtlas

qa-digestlisted

Watch and digest a local video file — transcribe its narration and export keyframes so Claude can actually read what happens. Use whenever someone points at a video on disk (.mp4, .mkv, .mov, .webm, .avi) and wants it watched, summarized, broken down, analyzed, or when a screen recording narrates a bug/feature/review out loud: "digest this movie", "watch this", "what's in this footage", "transcribe and break down this clip", or any bug-report / UI-review screen recording. Also files the bugs it finds as GitHub issues, with keyframes attached, when asked to "file these", "open tickets for these", or "turn this into issues". Local files only.
ShreddyKrueger75/qa-digest · ★ 1 · Code & Development · score 74
Install: claude install-skill ShreddyKrueger75/qa-digest
# qa-digest ## Introduction `qa-digest` is a local-only QA tool for narrated screen recordings. It transcribes the audio, extracts meaningful keyframes, and produces a timestamped digest so visual bugs can be verified against what was said. It does not upload recordings or file issues unless explicitly requested. Claude can't decode video or hear audio. This skill splits the job: a script does the mechanical part — transcribes the spoken audio with timestamps and exports one keyframe per scene — and then **Claude reads the transcript and views the frames** to write the digest. **The audio is the point.** A narrated screen recording (someone talking through a bug) is nearly useless as frames alone — the actual report is in the voice. Transcribe first, read `transcript.md`, *then* look at the frames. Skipping the transcript and pulling frames with raw `ffmpeg` is the #1 way to miss the whole message. ## Prerequisites ```bash brew install ffmpeg pip install Pillow numpy faster-whisper scenedetect ``` `ffmpeg`/`ffprobe` are required. `faster-whisper` and `scenedetect` are optional and auto-detected — without the first it skips the transcript; without the second it samples frames at even time intervals (fine for screen recordings). ## Setup (first run) On first run in an interactive terminal, the script prompts for four defaults: ``` (1) Diff-threshold mode — how aggressive frame selection is (2) Generate HTML report by default? (3) Output directory (4) Whisper model ```