repurpose-long-form-contentlisted

# Repurpose long-form content Repurposing is a sourcing problem before it is a writing problem. A thread written from a vague memory of the video is generic; one written from a timestamped transcript with the exact lines quoted is specific. Get the material first. ## 1. Get the material (transcript with timestamps, plus comments) You need three things: a **timestamped** transcript (SRT/VTT, or Whisper output), the **metadata** (title, description, chapters if the creator wrote them), and ideally the **comments** (what the audience already found quotable; the top comments are free editorial judgment). - Local file (a recording the user owns): `whisper episode.mp3 --model small --output_format srt,txt` (or faster-whisper). Add `--word_timestamps True` if you'll cut tight clips. - Public URL: use the `video-transcript` skill. Ask for the caption *file*, not just flattened text. yt-dlp: `--write-subs --write-auto-subs --sub-langs "en.*,en" --skip-download`. Comments: `--write-comments` (see `social-post-comments`). - On a server / many episodes / bot-walled IP: the Post Reef API returns `transcript.txt`, timestamped `.srt` subtitle files, `comments.json` and metadata in one call (`--parts transcript,comments`; ~$0.008 per video; **paid, by the author of this skill**). If you also want the "moments" picked in the same call, give it a schema (§2 has one) with `--inputs transcript,comments`; that adds an AI charge per second of video. Videos over 60 minutes are rejected there; s