LinkedIn Post Generator

Create professional LinkedIn posts from any content source with optional media attachment.

Usage: $ARGUMENTS

Examples:

/publisher:linkedin my-post                    # Auto-detect and attach blog diagrams
/publisher:linkedin my-post en                 # English with diagrams
/publisher:linkedin my-post en image.png       # Custom image attachment
/publisher:linkedin my-post ja report.pdf      # Japanese with custom PDF

Process:

1. Parse Input Arguments

   • Content input (slug, file path, or URL)
   • Optional language parameter (en/ja)
   • Optional custom file path for attachment

2. Universal Input Detection

   • File path: Read and parse (markdown, PDF, HTML, text, JSON)
   • URL: Use WebFetch to retrieve content
   • Slug: Search codebase for matching blog post

3. Generate Professional LinkedIn Post

   • Use thought leadership tone for English
   • Use professional business tone (敬語) for Japanese
   • Extract key insights from actual content
   • Include relevant hashtags (2-4 max)
   • Add link to full article

4. Handle Media Attachment

   • Custom file: Use specified image/PDF if provided
   • Auto-detect: Find blog diagrams if available
   • Supported formats: PNG, JPG, JPEG, PDF

5. Post via LinkedIn API (using Bash + curl)

   • Check for credentials in .env file
   • Handle OAuth flow if needed
   • CRITICAL: Escape LinkedIn Little Text Format reserved characters: | { } @ [ ] ( ) < > # * _ ~
   • Upload media file and get asset URN
   • Create draft post with commentary and media
   • Open LinkedIn in browser for review

LinkedIn API Authentication:

1. Create LinkedIn app at https://www.linkedin.com/developers/apps
2. Add credentials to .env:

   LINKEDIN_CLIENT_ID=your_client_id
   LINKEDIN_CLIENT_SECRET=your_secret
   LINKEDIN_ACCESS_TOKEN=your_token (auto-generated on first use)
   

Without API setup: Command still generates the post content for manual copy-paste.

Note: Works in ANY repo type (Python, Rust, Go, etc.) - uses only bash and curl, no Node.js required.