Auto-Repurpose Long-Form Content into Multi-Platform Social Drafts with n8n

Long text → webhook/RSS trigger → AI Agent generates multi-format content → code-level rule enforcement → Google Sheets queue → JSON response or Slack notification

Who

Content ops teams, indie creators, marketers who need to batch-produce social content from long articles

You get

One long article becomes platform-ready drafts for Twitter threads, LinkedIn posts, Instagram captions, newsletter blurbs and more, with brand voice applied and human review required before publishing

You need

n8n · google-sheets · gemini · slack

Steps

1. Configure webhook entry and validate content length

   Create a Webhook node in n8n, path set to repurpose, method POST, response mode set to responseNode. Add an Edit Fields (Set) node to extract body.content, body.title, body.url, and set origin to webhook. Then add an If node to check whether the content field has ≥60 words when split on whitespace; if not, route to Respond to Webhook to return a 400 JSON error without calling the model.

2. Configure RSS scheduled entry (optional)

   Add a Schedule Trigger node set to run daily at 08:00. Connect to an RSS Feed Read node with your blog RSS URL. A subsequent Code node filters for posts published within 24 hours, ≥60 words, max 3 articles, strips HTML tags, and outputs a unified data structure with origin set to feed.

3. Prepare brief and inject brand voice

   Add a Code node (Prepare Brief) that truncates overly long articles to 12,000 characters at a sentence boundary, recording whether truncation occurred and word count. Then add a Google Sheets node to read the first row of the Brand Voice tab, followed by a Code node (Apply Brand Voice) that attaches brand name, audience, tone, call-to-action, hashtags, banned words, etc. to each brief. If the sheet is empty, sensible defaults keep the workflow running.

4. AI Agent generates multi-platform drafts

   Add an AI Agent node (Repurposing Agent), connecting an OpenAI Chat Model (default gpt-4.1-mini, temperature 0.6) and a Structured Output Parser. Configure a fallback to the Google Gemini Chat Model. In the Agent's text field, enter the user prompt; in options.systemMessage, enter the system instruction requiring JSON output of summary, keyPoints, hookVariants, tweetThread, linkedinPost, instagramCaption, newsletterBlurb, and hashtags.

   Execution prompt · Source text。AI Agent node (Repurposing Agent) text field

   Copy prompt

   =Repurpose the content below.
   
   Brand: {{ $json.voice.brandName }}
   Audience: {{ $json.voice.audience }}
   Tone: {{ $json.voice.tone }}
   Call to action: {{ $json.voice.callToAction }}
   Banned words: {{ $json.voice.bannedWords.length ? $json.voice.bannedWords.join(', ') : 'none' }}
   Source title: {{ $json.sourceTitle }}
   Source link: {{ $json.sourceUrl || 'not provided' }}
   Source length: {{ $json.wordCount }} words{{ $json.truncated ? ' (trimmed for length, work with what is here)' : '' }}
   
   Content:
   {{ $json.content }}

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

   Copy prompt

   You turn one long piece of content into ready to post drafts for several channels.
   
   Voice
   - Write as the brand you are given, for the audience you are given, in the tone you are given.
   - Never use the banned words listed in the brief.
   - Never invent statistics, quotes, prices, dates or customer names. Use only what the source text says.
   - Do not use emoji, exclamation marks or phrases like game changer, unlock, supercharge or dive in.
   
   Per channel
   - summary: two or three sentences a busy reader can skim.
   - keyPoints: four to six short takeaways, each a full sentence.
   - hookVariants: three different opening lines for the same idea, so a human can pick one.
   - tweetThread: four to seven posts. Each one must be under 270 characters and must make sense on its own.
   - linkedinPost: one post of roughly 120 to 200 words, first person, ending in the call to action from the brief.
   - instagramCaption: under 150 words, plain sentences, no hashtags in the text itself.
   - newsletterBlurb: a short paragraph that teases the piece and links out.
   - hashtags: three to five topical tags without the hash symbol.
   
   Return only the requested JSON fields.

5. Enforce rules in code and queue drafts

   Add a Code node (Enforce Channel Limits) with hardcoded limits: tweet 280, linkedin 3000, instagram 2200, newsletter 900, summary 600. Performs banned-word cleaning, over-length trimming/splitting, hashtag deduplication and capping, and records every repair. Then a Code node (Split Into Channel Rows) splits each draft into one row per channel. A Code node (Build Run Summary) aggregates results. Finally, a Google Sheets node (Save Drafts to Queue) writes to Content Queue in appendOrUpdate mode, status set to Needs review.

6. Return JSON or send Slack notification

   Add an If node (Came From The API?) checking if origin equals webhook: if yes, route to Respond to Webhook returning the full JSON payload; otherwise route to Slack node to send a run summary to the configured channel.

You are setting up an n8n workflow that turns long text into multi-platform social drafts. Confirm in this order: 1) n8n instance is accessible with OpenAI + Google Sheets + Slack credentials; 2) Google Sheet created with Brand Voice and Content Queue tabs and correct column names; 3) Decide whether to use webhook entry, RSS scheduled entry, or both; 4) Import workflow.json or build nodes manually; 5) Replace Repurposing Agent text and systemMessage with the source prompts; 6) Test with sample content ≥60 words to verify JSON output format.