video-caption-creation_skill

Overview

This skill generates on-screen text hooks and captions for short-form video clips, built around the Complementarity Principle: on-screen text should add to the audio, not repeat it. It includes a podcast-clip workflow, hook categories, First-3-Words test, and Triple Word Score optimization to improve stop-rate and algorithmic relevance.

How this skill works

Scan the clip or transcript to find a hookable moment (surprising stat, provocative quote, emotional peak). Select a complementarity pattern (Question→Answer, Teaser→Payoff, Reframe→Evidence, etc.) and produce 3–5 short hook options (3–8 words). Run quality gates (First-3-Words, McDonald's Test, Gap Test) and mark one recommended pick. Then write a single short-form caption and an optional shorter X variant with handles and hashtags.

When to use it

• Creating on-screen overlays for podcast clips and short-form videos
• Optimizing clips from long-form episodes to increase watch-through and engagement
• Testing alternative hooks to A/B headline performance
• Batch-producing captions and hashtag sets for multi-platform posting
• When a clip contains a clear statistic, named framework, or emotional peak

Best practices

• Always apply the Complementarity Principle: add context, don’t repeat audio
• Front-load punch: evaluate the first three words for impact
• Generate 3–5 distinct hook styles and pick one recommended option
• Lead captions with the strongest line, include guest and 10–12 hashtags
• Use credential leads only when the guest or data justify authority

Example use cases

• Turn a 60–90s podcast moment with a surprising stat into four on-screen hook options and a caption ready for Shorts/Reels/TikTok
• Scan a full episode transcript to identify Tier-1 hookable moments and extract clips for editors
• Create rapid A/B variations for a clip: Polarizing, Curiosity Gap, Credential Lead
• Produce single-caption copy that aligns with on-screen text, transcript keywords, and a 10–12 hashtag strategy
• Apply Triple Word Score so topic words appear on-screen, in captions, and in transcript-first-10-seconds