Feed Claude Code a blog post, newsletter, or article from your project and get back platform-specific social media content — LinkedIn posts, Twitter/X threads, and Instagram carousel copy — written directly to your filesystem.
What Makes This a Claude Code Skill
Claude Code reads the source file from your project, understands the full context, and writes multiple output files to a social content directory. No copy-pasting between tabs.
Based on the real thread-writer and social-content skills from OpenClaudia/openclaudia-skills and coreyhaines31/marketingskills.
Setup
Create .claude/commands/repurpose.md in your project root:
Take the content file at the path provided by the user argument $ARGUMENTS and create platform-specific social media content.
## Process
1. Read the source file provided as an argument
2. Extract core insights, key data points, quotable lines, and the main narrative
3. Create the output directory: `./social-content/[slug-from-filename]/`
4. Generate and write the following files:
### LinkedIn (linkedin.md)
- 3 posts with different angles: story hook, contrarian hook, value hook
- Each 150-300 words, line breaks between paragraphs, end with a question or CTA
### Twitter/X Thread (twitter-thread.md)
- One thread of 5-8 tweets
- Tweet 1: Hook that stops the scroll (under 280 chars)
- Tweets 2-7: Key insights, one idea per tweet
- Final tweet: Summary + CTA with link back to original
### Instagram Carousel (instagram-carousel.md)
- 8-10 slide outline
- Slide 1: Bold headline hook
- Slides 2-9: One insight per slide (under 30 words)
- Final slide: CTA
- Include a caption with 20-25 hashtags
5. Print a summary showing all files created and word counts
Usage
/project:repurpose src/content/blog/my-latest-post.md
Claude reads the post, generates all social content, and saves it to ./social-content/my-latest-post/.
Tips
- Add a
brand-voice.mdfile to your.claude/directory so Claude maintains your tone - Run immediately after publishing a blog post to batch your social promotion
- Your source article should be at least 800 words for best results