whisper-extractlisted

# Whisper Extract Transcribe audio or video with Whisper, then produce a `.md` file containing an AI summary followed by the complete literal transcript. ## Prerequisite check Before doing anything else, verify that Whisper is installed: ```bash whisper --help > /dev/null 2>&1 && echo "OK" || echo "NOT FOUND" ``` If not found, tell the user to run: ```bash pipx install openai-whisper brew install ffmpeg # if ffmpeg is missing ``` Then stop and wait — do not proceed until Whisper is available. --- ## Step 1: Gather required information Ask the following in a **single message** if the user has not already provided them. Never ask more than once, and never ask for things already mentioned in the conversation. **Required:** - **Audio/video file path** — absolute or relative path to the file (mp3, mp4, wav, m4a, ogg, flac, webm, mkv, mov, mpeg, mpga, oga, wma). - **Language** — spoken language in the recording. Examples: `Spanish`, `English`, `Portuguese`. If unsure, say "auto-detect" and Whisper will figure it out (slower). - **Recording context** — a short description of what this is (e.g., "team meeting about Q3 roadmap", "interview with a candidate", "product demo call", "lecture on clean architecture"). This is used to write a better summary. **Optional (ask only if not obvious):** - **Whisper model** — default is `medium`. Options: `tiny` (fastest, less accurate), `base`, `small`, `medium` (recommended balance), `large-v3` (most accurate, ~3 GB download). Ask i