Multi Platform Content Repurposer With n8n And Claude

Most content teams publish a blog post and move on. The post gets one share on LinkedIn, maybe a tweet, and then lives in silence on the website for years. Repurposing is universally acknowledged as high-ROI — the same ideas packaged for different platforms and formats — but it never happens consistently because it takes another 2–3 hours per post that no one has.

This workflow makes repurposing automatic. The moment your team marks a Notion post as “Published,” Claude receives the full content and simultaneously writes platform-native versions: a LinkedIn post with proper hooks and formatting for that algorithm, a genuine 10-tweet thread where every tweet has standalone value (not just “here’s tweet 4 of 10: …”), and a newsletter intro that teases rather than summarizes. Each format follows explicit stylistic rules in the prompt — no generic AI-sounding content, no copy-pasted summaries.

Buffer handles the scheduling intelligence: LinkedIn posts at your account’s peak engagement time, the Twitter thread hits the next morning when your audience is online, and the email intro queues into your newsletter workflow. The entire process from publish to “all three formats scheduled” takes under 90 seconds and costs $0.02 in API fees.

Prerequisites

1. n8n instance (Cloud or self-hosted; the free tier of n8n Cloud handles this workflow comfortably)
2. Notion workspace with a blog posts database; minimum required properties: Title (title), Status (select with a “Published” option), Content (page body — this is where the post is written)
3. Anthropic API key with Claude 3.5 Sonnet access
4. Buffer account with connected social profiles — LinkedIn Page or Personal, Twitter/X account, and optionally your email platform connected (Buffer supports Mailchimp and others)
5. Buffer API key — found at buffer.com/developers

Setup Guide

1. Set up the Notion database with a Status select field. Add the options: Draft, Review, Published. Instruct your team to flip status to Published when a post is ready — this is the only trigger step that requires a human action.
2. Configure the Notion Webhook in n8n — use n8n’s Notion trigger with event type “Page Updated.” Add a filter: only proceed if Status equals “Published.” Without this filter, every page edit will trigger the workflow.
3. Fetch the full Notion page content — use the Notion “Get a Page” node with the page ID from the trigger, then use “Get Block Children” to retrieve all blocks. Use an n8n Code node to flatten block content into plain text: concatenate paragraph text, heading text, and list items.
4. Map prompt variables — in the Claude HTTP Request node, map: BLOG_TITLE from the Notion page title, BLOG_URL from a custom Notion property (you’ll add this after publishing), BLOG_CONTENT from the flattened text, and hardcode BRAND_VOICE and TARGET_AUDIENCE with your company’s specifics.
5. Parse the output — Claude returns all three formats in one response separated by ---. Use n8n’s Split Out node or a Code node with response.split('---') to extract each format into separate fields. Split the Twitter thread by newline to get individual tweet strings.
6. Configure Buffer API calls — three HTTP POST requests to https://api.bufferapp.com/1/updates/create.json:
   • LinkedIn: text = linkedInPost, profile_ids[] = your_linkedin_id, scheduled_at = now + 2 hours
   • Twitter Thread: use Buffer’s thread endpoint with the array of tweet strings
   • Email: if using Buffer’s email integration, POST to the newsletter draft endpoint
7. Test end-to-end — create a Notion test page, set status to Published, and verify all three Buffer drafts appear correctly. Check character counts on the LinkedIn post (Buffer will reject posts over the limit).
8. Add a Slack notification (optional but recommended) — after Buffer scheduling, send a Slack message to your marketing channel with links to review the drafts before they go live. This gives a human checkpoint without adding friction.

Who This Is NOT For

• Teams publishing fewer than 2 posts per month — the setup time isn’t justified below that cadence. Manual repurposing is faster.
• Highly technical or niche content (e.g., academic papers, medical content, legal analysis) — Claude may over-simplify complex technical content when adapting for social formats. Always have a subject matter expert review outputs.
• Brands with very strict tone-of-voice guidelines — Claude follows the BRAND_VOICE prompt variable, but if your brand voice is unusually specific (a particular named personality, proprietary vocabulary), you’ll need to extend the prompt significantly with examples.
• Teams that need LinkedIn and Twitter posts to go live immediately upon publishing — this workflow has built-in scheduling delays. If you need instant posting, remove the Buffer scheduling and post directly via the LinkedIn and Twitter APIs instead.