# The prompt

First give your agent the captions-dodge-hands skill (paste the skill text and ask it to save it as a skill).
Then replace everything in [BRACKETS] below and paste it in.

**Tip:** use any AI tools you like for the takes, the music and the effects, or none at all (your own phone videos
track best). If you'd rather not set up separate AI accounts and API keys, HYVIRAL (optional) is one way to get
video and music generation behind one connection (hyviral.com; in Claude.ai → Customize → Connectors → add
`https://api.hyviral.com/mcp`).

---

Use the captions-dodge-hands skill to make a vertical video for [PLATFORM: Instagram Reels / TikTok / YouTube Shorts].

- **The takes:** [FOOTAGE - e.g. "my own phone videos, attached, filmed on a tripod", or "an AI presenter made from
  this photo of me"]
- **Where it is filmed:** [PLACE - e.g. a laundromat with washer doors to bounce off and throw into, my kitchen with
  a fishbowl, my office window]
- **Who is talking:** [PRESENTER - e.g. me, a woman in her twenties, playful, relaxed American accent]
- **What I say:** [LINE - e.g. "use the skill's example lines", or my own script with one hand move per line]
- **The word I grab:** [GRAB WORD - e.g. ONE, from "Except this one"]
- **Length:** [LENGTH - e.g. about 23 seconds]
- **Language:** [LANGUAGE - e.g. English]
- **Music:** [MUSIC - e.g. playful funk at about 100 BPM with slap bass and clavinet, no vocals; or a track I own]
- **Look:** [STYLE - e.g. violet, coral and cyan chips with white heavy type, a gradient hand overlay]
- **AI tools:** [AI TOOLS - e.g. "whatever you already have set up", "HYVIRAL", or "Veo for the takes and Lyria for
  the music"]
- **Ending:** [CTA - e.g. "Comment DODGE and I'll send you the skill", or "Follow for more"]
- **Avoid:** [AVOID - e.g. no chips over my face, no brand logos]

Work in this order:

1. Write the line with one hand move per line on the music grid, and pick what the chips bounce off and what I throw
   the word into.
2. Build a storyboard page (per moment: a concept frame with the chips around my real hands, the line, the gesture,
   every sound, the music section, and the exact prompt for every generated asset). Show it to me and wait for my OK.
3. Make or check the takes (camera completely still, slow open-handed gestures), then track my hands and face on
   every frame and find the clap, the burst, the sweep, the flick, the grab and the toss from the landmarks.
4. Place the music so its downbeat lands on the burst, precompute the chip physics in Python, and turn every dodge,
   bounce and landing into a note in the song's key.
5. Build the dull bar, the shatter, the chips, the hand overlay, the ripples, the captions and the ending in Remotion,
   with the grabbed word on top of my hand, and mix voice first.
6. Run every QA gate in the skill (transcript, contact sheet, two identical renders, loudness, no gap over 1 second)
   and give me the MP4, a cover frame, and a ready-to-paste caption with hashtags.