content-repurposing

---
name: content-repurposing
description: Use when expanding one long-form video or essay into a multi-platform pack — Twitter thread, LinkedIn post, Reels/Shorts script, newsletter blurb — preserving voice and key beats from the source transcript.
source: https://www.anthropic.com/news/claude-3-5-sonnet
generated: 2026-05-07T21:42:15.312Z
category: concept
audience: creators
---

## Tutorials

- https://skillmake.xyz/v/content-repurposing.mp4

## When to use

- Multiplying one creator artifact (video, essay, podcast) into 4–6 platform-native posts
- Maintaining a single creator voice across Twitter, LinkedIn, Reels, and email
- Avoiding 'literally the same caption everywhere' which kills engagement
- Producing a publishable batch in one pass instead of hand-writing each variant

## Key concepts

### platform native voice

Each platform rewards a different shape: Twitter = punchy + numbered + ends with quote-tweet bait; LinkedIn = first-person story + soft CTA; Reels = hook in first 3 words + short sentences; Newsletter = warmer + section breaks + one big idea. The same source text rewritten for each.

### beat extraction

Pull 5–8 'beats' from the source — discrete claims, stories, numbers, or moments. Each beat becomes a tweet, a LinkedIn paragraph, or a Reel hook. The pack stays coherent because every variant covers the same beats in different shapes.

### voice anchoring

Pin a 200-word voice sample (an existing post by the creator) into every prompt as a style anchor. Without this, output drifts toward 'generic LinkedIn-speak' and feels off-brand. Voice sample > tone instructions every time.

## API reference

```
extract-beats prompt
```

First pass over the source transcript or essay. Returns JSON list of beats with type (claim/story/number/moment) and a 1-line summary each.

```
From the source below, extract 5–8 beats. For each: type (claim|story|number|moment), one-line summary, and 1–2 quotable lines verbatim. Return JSON.

SOURCE: <transcript>
```

```
platform-rewrite prompt with voice anchor
```

Second pass: per platform, generate the post(s) using the extracted beats and a voice sample.

```
STYLE ANCHOR (creator's existing post — match tone, cadence, sentence length):
<voice-sample>

BEATS:
<json-from-pass-1>

Write a 7-tweet thread for Twitter using these beats. Each tweet ≤220 chars. Open with the strongest hook beat. Close with a quote-tweet-bait question.
```

## Gotchas

- Don't try to write all 5 platforms in one prompt — output collapses to mush. Run platform-by-platform with the same beat list.
- Voice anchor must be the creator's own writing, not a 'pretend you're <famous person>' instruction.
- Reels/Shorts scripts must include scene directions ([cut], [zoom], [text overlay]); pure prose makes the script useless to the editor.
- LinkedIn truncates after ~210 chars before 'see more' — front-load the most interesting beat.

---
Generated by SkillMake from https://www.anthropic.com/news/claude-3-5-sonnet on 2026-05-07T21:42:15.312Z.
Verify against source before relying on details.

File: ~/.claude/skills/content-repurposing/SKILL.md