Give Claude Code Eyes: The /watch Skill That Lets Your Agent Watch Any Video

The gap Claude Code couldn't close

Claude Code can read a webpage, run a script, browse a repo, analyze an image. What it can't do natively is watch a video. Paste a YouTube link and it has to guess from the title, pull a transcript that misses half of what's on screen, or admit it can't help.

The

/watch

Installing the skill

bashCopy
# Claude Code (CLI)
/plugin marketplace add bradautomates/claude-video
/plugin install watch@claude-video

# Codex / generic skills runtime
git clone https://github.com/bradautomates/claude-video.git ~/.codex/skills/watch

For claude.ai (web): download

watch.skill

How to use it

• Analyze a video:
  promptCopy

  /watch https://youtu.be/dQw4w9WgXcQ what happens at the 30 second mark?

• Reverse-engineer content:
  promptCopy

  /watch https://youtu.be/abc123 what hook did they open with?

• Debug from a recording:
  promptCopy

  /watch bug-repro.mov what's going wrong?

• Summarize:
  promptCopy

  /watch https://youtu.be/xyz summarize this

• Focused window:
  promptCopy

  /watch https://youtu.be/abc --start 0:45 --end 1:15 what changed on screen?

What the skill actually does

When you call

/watch

1. promptCopy

   yt-dlp

   downloads the video into a temp directory (local files are probed in place)
2. promptCopy

   ffmpeg

   extracts frames at a duration-aware rate
3. Transcript comes from native captions via
   promptCopy

   yt-dlp

   (free), with Whisper API as fallback
4. Frames + transcript are handed to Claude, which reads each frame as an image in parallel
5. Claude answers grounded in what's on screen and in the audio

Zero config on first run —

yt-dlp

ffmpeg

brew

The frame budget — why it matters for token cost

Every frame is an image token. The skill's auto-fps logic exists to avoid blowing your context budget on a long video:

• ≤30 seconds: ~30 frames — dense, essentially every key moment
• 30s–1 min: ~40 frames — still dense
• 1–3 min: ~60 frames — comfortable
• 3–10 min: ~80 frames — sparse but workable
• >10 min: 100 frames — use
  promptCopy

  --start

  /
  promptCopy

  --end

  for focused questions

What people actually use it for

• Content analysis — break down a competitor's hook structure, pacing, and CTA from the actual video, not the title
• Bug reproduction — watch a screen recording of something broken and identify the frame where the issue appears
• Video summarization — structure and key moments faster than 2x speed
• Ad creative analysis — paste a TikTok or Reel URL and ask what made the first 3 seconds work

Frequently asked questions

What video sources does the skill support? Anything yt-dlp supports — YouTube, Loom, TikTok, X, Instagram, Vimeo, and hundreds of other platforms. Local

.mp4

.mov

.mkv

.webm

Do I need a Whisper API key? No, not for most videos. The skill first tries native captions via yt-dlp — free and instant. Whisper only kicks in for videos with no captions.

How do I get more detail on a specific moment in a long video? Use

--start

--end

/watch https://youtu.be/abc --start 2:30 --end 3:00 what's happening here?