| name | social-video |
| description | Create 30-second social videos from essay content using HeyGen Video Agent. Use when publishing content to generate shareable LinkedIn/Twitter videos with avatar, captions, and professional background. Requires HEYGEN_API_KEY in .zshrc. |
Social Video Skill
Generate short 30-second social videos from content using HeyGen Video Agent.
Prerequisites
HEYGEN_API_KEY environment variable (set in ~/.zshrc)
- Essay or content to convert
Video Configuration
| Setting | Value |
|---|
| Avatar | Master Control (suit): 4cadeeea56e14c05bb017756c98ca9a7 |
| Voice | Tom: 0356eda1a1414066bafac71d0f1e046c |
| Orientation | Portrait (9:16) for LinkedIn/social |
| Background | Professional brick wall studio |
| Captions | Enabled |
| Duration | ~30 seconds |
Content Structure
Each video follows a 4-scene structure:
| Scene | Duration | Purpose |
|---|
| Hook | 4-5 sec | Attention-grabbing opening |
| Problem | 8 sec | State the challenge |
| Solution | 12 sec | Key insight or approach |
| CTA | 5-6 sec | Actionable takeaway |
Example Script Structure
Scene 1 (Hook, 5 sec):
VO: "You decided AI is strategic. Now who gets it first?"
Scene 2 (Problem, 8 sec):
VO: "Most companies give licenses to everyone. That produces only 12% adoption."
Scene 3 (Solution, 12 sec):
VO: "The winners deploy deliberately. Customer support, engineering, finance. High-volume, repeatable work."
Scene 4 (CTA, 5 sec):
VO: "One team. Specific work. Start there."
API Usage
Create Video
curl -X POST "https://api.heygen.com/v1/video_agent/generate" \
-H "X-Api-Key: $HEYGEN_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"prompt": "Create a 30-second LinkedIn/social video about [TOPIC]...\n\nGlobal Style:\n- Professional brick wall studio background\n- Clean, minimal modern aesthetic\n- Portrait orientation (9:16) for social media\n- Enable captions\n\nScene Structure:\n\nScene 1 (Hook, 5 sec):\nVO: \"[hook text]\"\n\nScene 2 (Problem, 8 sec):\nVO: \"[problem text]\"\n\nScene 3 (Solution, 12 sec):\nVO: \"[solution text]\"\n\nScene 4 (CTA, 5 sec):\nVO: \"[call to action]\"",
"config": {
"duration_sec": 30,
"avatar_id": "4cadeeea56e14c05bb017756c98ca9a7",
"orientation": "portrait"
}
}'
Check Video Status
curl -s "https://api.heygen.com/v1/video_status.get?video_id=VIDEO_ID" \
-H "X-Api-Key: $HEYGEN_API_KEY" | jq '.'
Poll Until Complete
Status progression: pending → processing → completed
Poll every 60-90 seconds until status: "completed".
Download Video
When complete, the response includes:
video_url — Main video
video_url_caption — Video with burned-in captions (recommended for social)
gif_url — Animated GIF preview
thumbnail_url — Static thumbnail
curl -o output-captioned.mp4 "$VIDEO_URL_CAPTION"
Workflow
- Extract Key Points: From essay content, identify hook, problem, solution, CTA
- Write Script: Keep total duration ~30 seconds (count ~150 words max)
- Generate Video: Submit to HeyGen Video Agent API
- Poll for Completion: Wait for
status: "completed"
- Download: Get captioned video for social sharing
- Share: Upload to LinkedIn, Twitter, etc.
Script Writing Guidelines
Hook (4-5 seconds)
- Start with a question or counterintuitive statement
- One clear sentence
- Make reader want to know more
Problem (8 seconds)
- State the challenge clearly
- Use specific numbers when possible
- "Most companies..." or "The problem is..." patterns work well
Solution (12 seconds)
- Give the key insight
- List 2-3 specific examples or steps
- Keep practical and actionable
CTA (5-6 seconds)
- One clear action
- End with momentum, not "thanks for watching"
Storage
Videos are stored in: ~/.openclaw/clu-space/artifacts/heygen-videos/
Naming convention: {content-slug}-captioned.mp4
Example: From Essay to Video
Essay excerpt:
"Most companies get deployment wrong by giving licenses to everyone. That produces 12% adoption. The winners deploy deliberately to customer support, engineering, finance — high-volume, repeatable work. One team. Specific work. Start there."
Video script:
Scene 1: "Where do you start with AI?"
Scene 2: "Most companies give everyone licenses. That produces 12% adoption."
Scene 3: "The winners deploy deliberately. Customer support, engineering, finance. High-volume, repeatable work."
Scene 4: "One team. Specific work. Start there."
Notes
- Video Agent typically takes 2-5 minutes to generate
- Portrait (9:16) is optimized for LinkedIn/mobile
- Captions are burned into the video, not separate
- Background is HeyGen's built-in brick wall studio
- Avatar uses natural gestures and expressions