SuperAPI AI Directory

video-generation-skill

Design & creation

View repository

Updated 10-16 03:12

About this skill

Video Generation Skill

Purpose

Translate marketing goals and decision moments into concrete video plans that can be executed by external tools (VEO, Gemini, Nano Banana 2, etc.).

Outputs

Per video:

  • Hook (first 3–5 seconds)
  • Full script / outline
  • Suggested B-roll and overlay text
  • Scene-by-scene breakdown
  • Transitions and on-screen prompts
  • Thumbnail concept and text

Video Types

Short-Form (15-60 seconds)

  • TikTok
  • Instagram Reels
  • YouTube Shorts
  • Facebook Reels

Structure:

  1. Hook (0-3s): Pattern interrupt
  2. Problem (3-10s): Identify pain
  3. Solution (10-45s): Show the fix
  4. CTA (45-60s): What to do next

Long-Form (2-10 minutes)

  • YouTube
  • LinkedIn Video
  • Website embeds

Structure:

  1. Hook (0-15s)
  2. Introduction (15-30s)
  3. Main content (30s-8m)
  4. Summary (8-9m)
  5. CTA (9-10m)

Platform Guidelines

YouTube

  • Thumbnail: 1280x720, high contrast, max 3 words
  • Title: 40-60 chars, keyword-first
  • Description: First 150 chars are preview

TikTok

  • Vertical 9:16
  • Native captions required
  • Trending sounds boost reach

Instagram

  • Square or vertical
  • Auto-captions on
  • Cross-post to Facebook

LinkedIn

  • Professional tone
  • Captions essential (most watch muted)
  • Value-first approach

Usage

  • Called by SocialPlaybookAgent or VisualExperienceAgent
  • Use routes and components already defined in Synthex for embedding video demos
  • Store concepts in social_assets table with asset_type='video'