From latent-space-youtube
This skill should be used when the user asks to "upload Zoom recordings to YouTube", "publish Zoom videos", "download and upload recordings", or wants to run the full Zoom-to-YouTube pipeline. It orchestrates the zoom-download, youtube-publish, and youtube-thumbnails sub-skills with checkpoint stages between each phase.
How this skill is triggered — by the user, by Claude, or both
Slash command
/latent-space-youtube:zoom-to-youtubeThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
End-to-end workflow to download Zoom cloud recordings, upload them to YouTube with proper metadata, and generate custom thumbnails. This is an **orchestrator skill** — each stage is handled by a focused sub-skill that can also be used independently.
End-to-end workflow to download Zoom cloud recordings, upload them to YouTube with proper metadata, and generate custom thumbnails. This is an orchestrator skill — each stage is handled by a focused sub-skill that can also be used independently.
The pipeline has checkpoint stages between each action phase. At each checkpoint, gather context, present a summary to the user, and get confirmation before proceeding. This catches mismatches, duplicates, and edge cases early.
Goal: Understand the current state of both Zoom and YouTube before doing anything.
zoom.us/recording (log in via Google SSO if needed)Compare the two lists and identify:
→ CHECKPOINT: Present the user a table like this and ask for confirmation:
Zoom recordings found:
1. AI in Action Weekly Jam! — 7 Feb 2026 — 53:46 — NEW, will download
2. AI in Action Weekly Jam! — 31 Jan 2026 — 58:45 — Already on YouTube, skip
3. Johan Duramy and swyx — 12 Feb 2026 — 56:22 — NEW (Paper Club), will download
Flags:
- None
Proceed with downloading #1 and #3?
Wait for user confirmation. They may say "skip #3" or "also grab the Jan 31 one" etc.
Sub-skill: zoom-download
Only download the recordings confirmed in Stage 0. For each:
_gallery_ (not _gvo_)→ CHECKPOINT: Present download results and frame analysis:
Downloaded 2 recordings:
1. Feb 7 — SpaceMolt demo by Ian Langworth (statico)
- 53:46, screen share shows space game UI
- Participants: Ian, swyx, Sebastian, Ted
- Type: AI in Action (weekly jam demo)
2. Feb 12 — SDPO paper reading by Ted Kyi
- 56:22, screen share shows arxiv paper
- Paper: Self-Distillation Preference Optimization
- Participants: Ted, Johan, swyx
- Type: Paper Club
Proposed titles:
1. "SpaceMolt - AI Agents in Multiplayer Space Games: AI in Action 7 Feb 2026"
2. "RL via Self-Distillation (SDPO) Paper Reading: AI in Action 12 Feb 2026"
Proposed playlists:
1. → AI in Action
2. → Paper Club
Any changes to titles or classifications?
Wait for user feedback. They might correct a presenter name, adjust the title wording, or reclassify a video.
Sub-skill: youtube-publish
→ CHECKPOINT: Present published video summary:
Published 2 videos:
1. "SpaceMolt - AI Agents in Multiplayer Space Games: AI in Action 7 Feb 2026"
- URL: https://youtu.be/XXXXX
- Playlist: AI in Action
- Status: Unlisted ✓
2. "RL via Self-Distillation (SDPO) Paper Reading: AI in Action 12 Feb 2026"
- URL: https://youtu.be/YYYYY
- Playlist: Paper Club
- Status: Unlisted ✓
Ready to generate thumbnails?
Sub-skill: youtube-thumbnails
→ CHECKPOINT: Confirm thumbnails are set:
Thumbnails set:
1. SpaceMolt — space-themed, neon ships & asteroids ✓
2. SDPO — academic brain/neural net with formulas ✓
All done! Anything to adjust?
| Skill | Use Standalone When |
|---|---|
zoom-download | Just need to grab recordings, no YouTube work |
youtube-publish | Videos already downloaded, just need to upload & title |
youtube-thumbnails | Videos already published, just need thumbnails |
_gvo_ = gallery view only (no screen share). Always use _gallery_.convert -resize 1280x720 -quality 85.npx claudepluginhub swyxio/claude-plugins --plugin latent-space-youtubeManages YouTube channels via CLI: upload videos, list/edit metadata, schedule/unschedule, set thumbnails, download private videos, fetch transcripts, generate AI chapters. Use for YouTube upload, scheduling, editing, downloading, transcribing.
Processes videos: download, trim, silence removal, audio extraction, transcription (Whisper), captions, speed change, concatenation, audio enhancement, upload to YouTube/Bunny.net, social posts. Requires ffmpeg, yt-dlp, and API keys.
Downloads videos from Google Drive, creates title slides, stitches with FFmpeg, and generates timestamped YouTube descriptions for course video assembly.