Difficulty: Intermediate Time to set up: 20 minutes Time per repurpose run: 10 minutes

Turn one blog post into 10 LinkedIn posts, 5 tweet threads, 3 email drafts, and a carousel outline. One input, many outputs.

---

Why This Matters

Kevin Kerner wrote on his Substack that he "cancelled a part-time marketing hire because of it." He was referring to a Claude Code content system that takes one piece of long-form content and turns it into a full week of multi-channel posts. That is not hyperbole. When your repurposing system is built right, a single blog post genuinely produces 20+ pieces of content, each tailored to the platform it lives on.

The trick is that repurposing is not copy-pasting. A LinkedIn post has different structure than a tweet thread. An email has a different job than a carousel. What makes this system work is that each output format has its own rules, and Claude Code follows all of them from a single skill file.

What You Will Build

A repurposing skill that takes any long-form content as input and produces platform-specific outputs for LinkedIn, Twitter/X, email, and carousel formats. Each output follows the rules of its platform.

Tools used: Claude Code (terminal), CLAUDE.md, skill files

---

Step 1: Create the Repurpose Skill File

Create skills/repurpose.md:

# Content Repurposing Engine

## Purpose
Take one piece of long-form content (blog post, article, transcript)
and produce multiple platform-specific outputs. Each output should
stand alone — someone reading just the LinkedIn post should get
value without needing to read the original.

## Input
The user will provide one of:
- A blog post (pasted text or URL to fetch)
- An article or newsletter
- A video transcript
- A podcast transcript

## Step 1: Extract Core Ideas
Before generating any output, extract:
- Main thesis (1 sentence)
- 3-5 key insights or claims
- Any data points, stats, or specific results
- Quotable lines or strong opinions
- The "so what" — why should the audience care

## Output Formats

### LinkedIn Posts (generate 10)
Rules:
- Max 1,300 characters each
- Each post focuses on ONE idea from the source
- Hook on line 1 (bold claim, stat, or question)
- Short paragraphs, one thought per line
- End with a discussion question
- No hashtags in body, max 3 at the very end
- No links in the post body
- Each post should have a different angle:
  1. The contrarian take
  2. The "here's what I learned" story
  3. The numbered list/framework
  4. The mistake/failure lesson
  5. The prediction or hot take
  6. The case study summary
  7. The "most people think X, actually Y" post
  8. The practical tip post
  9. The question post (engage the audience)
  10. The data/stat-driven post

### Tweet Threads (generate 5)
Rules:
- Each thread is 4-8 tweets
- Tweet 1 is the hook (must work standalone)
- One idea per tweet, max 280 characters each
- Use "Thread:" or a hook format on tweet 1
- Last tweet: summary + CTA (follow, retweet, link)
- No hashtags except optionally on the last tweet
- Number format: "1/ Here's what happened..."
- Each thread should cover a different key insight

### Email Drafts (generate 3)
Rules:
- Subject line (max 50 characters, curiosity-driven)
- Preview text (max 90 characters)
- Body: 200-400 words
- Conversational tone, like writing to one person
- One clear CTA per email
- Format options:
  1. The "one big idea" email
  2. The "3 things I learned" roundup
  3. The "story + lesson" narrative

### Carousel Outline (generate 1)
Rules:
- 8-12 slides
- Slide 1: Hook/title slide (bold statement or question)
- Slides 2-10: One idea per slide, max 30 words per slide
- Second-to-last slide: Summary or key takeaway
- Last slide: CTA (follow, visit, download)
- Include speaker notes for each slide (what to expand on)
- Format as numbered list with slide content + notes

## Quality Rules
- Every piece must stand alone (no "as I mentioned in my blog post")
- Adapt the voice to the platform (LinkedIn is more professional,
  Twitter is more punchy, email is more personal)
- Do not repeat the same hook across formats
- Include at least one specific detail (number, example, name) per piece
- If the source content is thin on a topic, skip that angle
  rather than generating filler

Step 2: Set Up Channel-Specific Rules

You can either keep everything in one file (above) or split into separate channel files for more detail. For most people, the single file works fine. If you find yourself tweaking one platform's rules often, break it out:

skills/
  repurpose.md          (main engine)
  repurpose-linkedin.md (optional: detailed LinkedIn rules)
  repurpose-twitter.md  (optional: detailed Twitter rules)
  repurpose-email.md    (optional: detailed email rules)

The main skill file references the others if they exist:

## Channel Detail Files (if present)
Check for platform-specific files in the skills folder.
If they exist, use those rules. If not, use the rules above.

Step 3: Connect It to Your CLAUDE.md

Add to your CLAUDE.md:

## Skill Files
- `skills/repurpose.md` — Content repurposing engine (blog > multi-channel)

And add a section about your channels:

## My Content Channels
- **LinkedIn:** [your profile URL] — primary thought leadership channel
- **Twitter/X:** [your handle] — shorter takes, engagement
- **Email list:** [approximate size] — [newsletter name/cadence]
- **Carousel posts:** LinkedIn carousels, designed in Canva

This context helps Claude Code tailor the outputs to your actual audience on each platform.

Step 4: Run It on a Blog Post

Start Claude Code and give it content to repurpose:

Repurpose this blog post into all channel formats:

[Paste your blog post here]

Or if the blog post is online:

Fetch this blog post and repurpose it into all channel formats:
https://yourblog.com/your-post-title

Claude Code will read the repurpose skill, extract the core ideas, and generate all outputs in one pass.

For long blog posts (2,000+ words), you will get richer outputs because there are more ideas to pull from. For shorter posts, you might get 6-7 LinkedIn posts instead of 10, which is fine. Quality beats quantity.

Step 5: Review and Schedule

The output will be long. Here is how to work through it efficiently:

First pass: Scan and star. Read through all outputs quickly. Mark the ones that feel strong immediately. These go first in your publishing queue.

Second pass: Edit the best ones. Take your top 5-7 pieces and do a quick edit. Read each one out loud. Tighten any sentences that feel flabby. Add personal details the AI could not have known.

Third pass: Schedule. Map your content to your publishing calendar:

Week 1:
  Mon: LinkedIn post #1 (contrarian take)
  Tue: Tweet thread #1
  Wed: LinkedIn post #3 (framework)
  Thu: Email #1 (one big idea)
  Fri: LinkedIn post #5 (hot take)

Week 2:
  Mon: LinkedIn carousel
  Tue: Tweet thread #2
  Wed: LinkedIn post #7 (most people think X)
  Thu: LinkedIn post #9 (question post)
  Fri: Email #2 (3 things I learned)

One blog post just gave you two weeks of content across three platforms.

Prompt Variations

You do not always need the full repurpose. Here are targeted prompts:

Repurpose this into LinkedIn posts only. Give me 5 posts,
each with a different angle.

[paste content]

Take this blog post and write 3 email drafts for my newsletter.
My list is mostly marketing directors at B2B companies.

[paste content]

Turn this podcast transcript into a tweet thread and a LinkedIn post.
Focus on the most surprising insight.

[paste content]

I published this LinkedIn post and it did well. Now turn it into
a longer email and a tweet thread that covers the same idea
from a different angle.

[paste post]

Making It Better Over Time

After a month of using the system, you will notice patterns. Some output types consistently need heavy editing. Others are nearly publish-ready. Update your skill file based on what you learn:

• If LinkedIn posts are too long, lower the character limit in your rules
• If tweets feel generic, add more examples of your Twitter voice
• If emails lack personality, add a section about your email tone specifically
• If carousels are too text-heavy, reduce the words-per-slide limit

The system gets better every time you refine the rules. That cancelled marketing hire Kevin Kerner mentioned? It did not happen on day one. It happened after a few weeks of tuning the system until the output quality was consistently high enough to trust.

File Structure

marketing-system/
  CLAUDE.md
  skills/
    repurpose.md
  content/
    blog-posts/
      2026-02-27-ai-visibility.md
    repurposed/
      2026-02-27-ai-visibility-linkedin.md
      2026-02-27-ai-visibility-tweets.md
      2026-02-27-ai-visibility-emails.md
      2026-02-27-ai-visibility-carousel.md

Keep your repurposed outputs organized by source and date. This makes it easy to find content later and track what you have already published from each source.

---

Tools used in this playbook: Claude Code CLI, web fetch (for URLs), CLAUDE.md, skill files, markdown