videolisted

# video: from a link to notes in the vault, and to frames when the text is blind This skill **extracts the text and the picture and keeps the books**; the quality of the notes belongs to the `notes` skill, called at step B. ``` link --> [0] duplicate? --> [A] transcript --> [B] notes writes the notes --> [C] registry + cleanup | +- text blind to the screen? --> [W] frames of the piece "look / what is on screen / cut a piece" -------------------------------------> [W] directly ``` - Vault: `${user_config.vault_dir}` (written `<vault>` below and in the references) - Scripts: `${CLAUDE_PLUGIN_ROOT}/skills/video/scripts` (written `<scripts>` below): `scripts/transcript.py` (step A), `scripts/frames.py` (step W), `scripts/check_done.py` (step C) - Notes language: ${user_config.notes_language} The scripts declare their own dependencies; always run them with `uv run`, which installs them on first use. A plain `python` lacks `yt_dlp` and `youtube_transcript_api`, and every tier drops out. ## Is this the right skill A video link can mean two tasks. This skill is for **knowledge from the video**: a lecture, a talk, a tutorial, a podcast interview, a bare link. If the user wants to understand **how the video is made** (hooks, pacing, editing of a clip, an ad, a vlog), that is a different task: say so and ask. When unsure, ask in one line: "do you want the knowledge from it, or how i