Auto-Generate LinkedIn Promo Posts from Ghost Blog with n8n + GPT-4o

Ghost CMS articles → HTML cleanup → GPT-4o LinkedIn copy → Google Sheets archive

Who

Content creators, bloggers, or marketing teams running a Ghost blog who want automated LinkedIn promotion

You get

Pull latest blog posts, clean HTML, auto-generate LinkedIn posts with hook, summary, CTA and signature, then log everything to Google Sheets

You need

n8n · Ghost CMS · OpenAI GPT-4o / GPT-4o-mini · Google Sheets · Code node

Steps

1. Configure Ghost and Google Sheets credentials

   After importing the template into n8n, set up two credentials: 1) Ghost node: add Ghost Blog Account Credentials and select the number of posts to fetch (default limit=3). 2) Google Sheets node: add Google Sheets API OAuth2 credentials, choose the target file and sheet, and confirm field mapping (id, title, content, excerpt, clean_content, linkedin_post, featured_image).

2. Extract and map blog post fields

   The Ghost node (Extract Blog Posts) pulls the post list in getAll mode. The Edit Fields (Extract Post Content) node then maps each post to: id, title, featured_image, excerpt, content (raw html), and link (url). These fields feed the cleanup and AI generation steps.

   Execution prompt · Source text。Edit Fields node assignments: id field expression

   Copy prompt

   ={{ $json.id }}

   Execution prompt · Source text。Edit Fields node assignments: title field expression

   Copy prompt

   ={{ $json.title }}

   Execution prompt · Source text。Edit Fields node assignments: featured_image field expression

   Copy prompt

   ={{ $json.feature_image }}

   Execution prompt · Source text。Edit Fields node assignments: excerpt field expression

   Copy prompt

   ={{ $json.excerpt }}

   Execution prompt · Source text。Edit Fields node assignments: content field expression

   Copy prompt

   ={{ $json.html }}

   Execution prompt · Source text。Edit Fields node assignments: link field expression

   Copy prompt

   ={{ $json.url }}

3. Clean HTML and feed to AI Agent

   The Code node (Clean HTML) runs JavaScript to strip HTML tags, normalize whitespace, and decode  , outputting clean_content. Then Loop Over Posts processes each item: title, link, and clean_content are injected into the AI Agent's user message, and GPT-4o-mini generates the LinkedIn promo copy according to the systemMessage structure.

   Execution prompt · Source text。Clean HTML Code node jsCode field

   Copy prompt

   const htmlContent = $input.first().json.content;
   
   const cleanText = htmlContent
     .replace(/<[^>]*>/g, '') // remove  tags
     .replace(/\s+/g, ' ')    // normalize spaces
     .replace(/&nbsp;/g, ' ') // decode common entity
     .trim();
   
   return [
     {
       json: {
         clean_content: cleanText
       }
     }
   ];

   Execution prompt · Source text。AI Agent node text field (user message template)

   Copy prompt

   =Article Title: {{ $json.title }}
   Article Link: {{ $json.link }}
   Article Content: {{ $json.clean_content }}

   Execution prompt · Source text。AI Agent node options.systemMessage field

   Copy prompt

   =You are a content marketing assistant. Based on the article metadata (ID, title) and cleaned content, generate a short LinkedIn promotional message for a professional audience.
   
   Follow this structure:
   
   Start with a hook that grabs attention (a bold insight, surprising fact, or thought-provoking question).
   
   Briefly summarize the article’s value — what readers will learn or gain from it.
   
   Include a clear call-to-action encouraging readers to read the article.
   
   End with this author signature and invitation:
   “—
   Samir Saci
   Supply Chain Data Scientist & Founder of LogiGreen
   📩 Contact me: https://logi-green.com/contactus”
   
   Use a professional and engaging tone. Do not include hashtags or Markdown formatting.

4. Merge results and write to Google Sheets

   The AI Agent output is merged with original post data via the Merge Linkedin node, then the Google Sheets node appends a row to the specified sheet in append mode. Written fields: id, title, content, excerpt, clean_content, linkedin_post (AI output), featured_image. After running, review the generated LinkedIn copy alongside source data in Google Sheets.

   Execution prompt · Source text。Google Sheets node columns.value: id

   Copy prompt

   ={{ $json.id }}

   Execution prompt · Source text。Google Sheets node columns.value: title

   Copy prompt

   ={{ $json.title }}

   Execution prompt · Source text。Google Sheets node columns.value: content

   Copy prompt

   ={{ $json.content }}

   Execution prompt · Source text。Google Sheets node columns.value: excerpt

   Copy prompt

   ={{ $json.excerpt }}

   Execution prompt · Source text。Google Sheets node columns.value: clean_content

   Copy prompt

   ={{ $json.clean_content }}

   Execution prompt · Source text。Google Sheets node columns.value: linkedin_post

   Copy prompt

   ={{ $json.output }}

   Execution prompt · Source text。Google Sheets node columns.value: featured_image

   Copy prompt

   ={{ $json.featured_image }}

You are a content marketing assistant. Based on the article metadata (ID, title) and cleaned content, generate a short LinkedIn promotional message for a professional audience.

Follow this structure:
1. Start with a hook that grabs attention (a bold insight, surprising fact, or thought-provoking question)
2. Briefly summarize the article's value — what readers will learn or gain from it
3. Include a clear call-to-action encouraging readers to read the article
4. End with an author signature and invitation

Use a professional and engaging tone. Do not include hashtags or Markdown formatting.