Use this skill whenever the user wants to write, plan, or refine a script for a talking-head video. Trigger on phrases like "write a talking-head script", "DTC video", "write a reel script", "script for me to film", "write what I should say", "talking head", "selfie video script", "I want to record myself talking", or whenever the user is producing a video where they (or a UGC creator) speak to camera. Direct-to-camera is the highest-converting format for new followers because the viewer is buying personality, not just information. This skill produces a 30 to 90 second script with hook, value beats, re-hooks, and CTA, formatted for one-line-at-a-time recording.
How this skill is triggered — by the user, by Claude, or both
Slash command
/hiruno-content-skills:direct-to-camera-scriptThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Brand context: !`cat .agents/brand-content-context.md 2>/dev/null || echo "No brand context found. Run /brand-content-context first to set up."`
Brand context: !cat .agents/brand-content-context.md 2>/dev/null || echo "No brand context found. Run /brand-content-context first to set up."
You write talking-head scripts that hit on the three format parameters: strong hook, real value, re-hooks. You also format the script for the production reality, which is that nobody actually memorises lines. Creators record one line at a time and cut the dead space.
Brand context: Read .agents/brand-content-context.md first. Voice, audience symptoms, banned words. Without context the script will sound like a generic creator. Tell the user to run /brand-content-context first if missing.
Direct-to-camera is the highest-converting format for new followers. The viewer hears personality and energy, not just words. They follow the person, not the topic. That's why this format gets the most follows per video even though it doesn't always get the most views.
It's also the format that punishes weak content most. There's nothing to hide behind — no B-roll, no slide design. The script and the delivery have to carry the post.
Every direct-to-camera script has four sections:
Total length: 30 to 90 seconds. Default to 45 to 60 unless the user specifies.
Direct-to-camera lives or dies by re-hooks. Without them, the viewer drifts at the first slow beat. Use these re-hook moves at retention-critical points:
Re-hooks are not filler. Each one should also be specific. "Here's the part most people miss" is filler. "Here's why every product person you've worked with quit by month four" is a re-hook.
Most creators don't memorise scripts. They read one line, look at the camera, deliver it, then move to the next. Format the script accordingly:
--- between linesThis is what the creator will actually look at while filming. Make it usable.
DIRECT-TO-CAMERA SCRIPT
Length: [estimated seconds]
Hook type: [question / list / story / negative / contrarian]
---
HOOK (0-3s)
[line]
---
SETUP (3-10s)
[line]
[line]
---
VALUE BEAT 1 (10-25s)
[re-hook line if applicable]
[line]
[line]
---
VALUE BEAT 2 (25-40s)
[re-hook line if applicable]
[line]
[line]
---
VALUE BEAT 3 (optional, 40-55s)
[re-hook line if applicable]
[line]
[line]
---
PAYOFF / CTA (last 5-8s)
[line]
---
PRODUCTION NOTES
- Frame: [phone vertical, head-and-shoulders / hands visible / etc.]
- Energy: [match the brand voice — calm, hyped, dry, etc.]
- Camera trick: record one line at a time. Read line, look up, deliver, cut.
- Editing: cut all dead space between lines. Add captions. Music low if any.
- Caption suggestion (Instagram): [1-2 sentence caption with relevant hashtags or none]
WHY THIS HOOK
[1-2 lines: which audience pain or desire it pulls from, which hook type, why it should pass the curiosity gap test]
If the user already has a hook (e.g. from hook-generator), use it. Build the rest of the script around it.
If they don't, ask which hook type they want, or pick based on:
If unsure, lean story or contrarian for personal-brand creators, negative or numbered list for product-led brands.
If the script is for a UGC creator filming for a brand (Cult of Sauna, Volt Dropper, etc.), add:
If the user gives you a draft and asks for fixes, look for these:
| Weakness | Fix |
|---|---|
| Hook is buried 5+ seconds in | Move the hook to line 1, cut the warm-up |
| Generic opening ("hey guys", "today I want to talk about") | Delete and replace with hook |
| No re-hook between beats | Add a numbered call-out or contrast pair |
| Long sentences (15+ words spoken) | Break into 2 lines, max 12 words each |
| Vague payoff ("hope this helps", "follow for more") | Replace with specific takeaway or specific CTA |
| Voice slips | Re-read brand sample paragraph, rewrite the off-voice lines |
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub angelobaricante/hiruno-content-skills --plugin hiruno-content-skills